Microsoft Azure AI Engineer Associate AI-102 (AI-102) — Questions 751825

988 questions total · 14pages · All types, answers revealed

Page 10

Page 11 of 14

Page 12
751
MCQhard

Your company uses Azure OpenAI Service to generate product descriptions. You need to reduce costs while maintaining response quality for common requests. What should you implement?

A.Use a smaller model variant (e.g., GPT-3.5 instead of GPT-4).
B.Batch multiple requests into a single API call.
C.Enable semantic caching to reuse responses for similar prompts.
D.Reduce the max_tokens parameter for all requests.
AnswerC

Caching reduces costs without sacrificing quality for common requests.

Why this answer

Semantic caching stores responses for semantically similar prompts, allowing Azure OpenAI Service to return cached results for common requests without incurring per-token inference costs. This directly reduces costs while preserving response quality, as the cached responses are identical to what the model would generate. Unlike other options, it avoids degrading output quality or requiring architectural changes.

Exam trap

The trap here is that candidates confuse semantic caching with simple request caching or assume batching is supported, overlooking that semantic caching specifically reuses responses for similar prompts without degrading quality, while other options either reduce quality or are technically infeasible.

How to eliminate wrong answers

Option A is wrong because using a smaller model variant (e.g., GPT-3.5 instead of GPT-4) reduces response quality and capability, contradicting the requirement to maintain response quality. Option B is wrong because Azure OpenAI Service does not support batching multiple requests into a single API call; each request is processed independently, and batching would require custom orchestration without cost savings. Option D is wrong because reducing max_tokens for all requests truncates responses, degrading quality for common requests that may require longer outputs, and does not address cost reduction for repeated or similar prompts.

752
MCQhard

You are using Azure AI Language to analyze medical records. The built-in NER model does not recognize some medical terms. What should you do?

A.Use the Text Analytics for Health feature
B.Use Entity Linking
C.Use Custom Text Classification
D.Train a custom NER model using Azure AI Language
AnswerD

Custom NER can be trained on your specific medical terms.

Why this answer

Custom NER allows training a model on your specific terms. Text Analytics for Health is a specialized built-in model for healthcare but may not cover all custom terms. Custom Text Classification is for document classification.

Entity Linking disambiguates to a knowledge base.

753
MCQhard

You are deploying a Custom Vision model to a production environment. The model must handle 100 predictions per second with low latency. Which deployment option should you choose?

A.Use the Free tier prediction endpoint.
B.Export the model as a Docker container and run it on Azure Container Instances.
C.Use the Training API to make predictions.
D.Use a paid tier prediction endpoint with sufficient capacity.
AnswerD

Provides dedicated resources for high throughput.

Why this answer

Option B is correct because a dedicated prediction endpoint with a paid tier ensures sufficient throughput and low latency. Option A is wrong because the Free tier has very limited transactions per second. Option C is wrong because the Export feature exports the model for offline use, not for real-time cloud serving.

Option D is wrong because the Training API is for training, not prediction.

754
MCQmedium

Refer to the exhibit. You have an Azure AI Language resource named MyLangService. You need to call the conversational language understanding (CLU) API. Which URL should you use?

A.https://mylangservice.cognitiveservices.azure.com/text/analytics/v3.0/sentiment
B.https://mylangservice.cognitiveservices.azure.com/language/analyze-conversations?api-version=2022-10-01-preview
C.https://mylangservice.cognitiveservices.azure.com/language/:analyze-conversations?api-version=2022-10-01-preview
D.https://mylangservice.cognitiveservices.azure.com/
AnswerC

This is the exact CLU endpoint shown in the exhibit.

Why this answer

Option A is correct because the CLU endpoint is shown as the full URL with the path '/language/:analyze-conversations'. Option B is wrong because that URL is for Text Analytics. Option C is wrong because it uses the Text Analytics base URL incorrectly.

Option D is wrong because it uses an invalid path.

755
MCQmedium

Your company uses Microsoft 365 Copilot to generate meeting summaries. Some users report that summaries include information from meetings they did not attend. What is the most likely cause?

A.The meeting organizer granted everyone in the organization view access.
B.Users have access to meeting artifacts via shared calendars or transcripts.
C.Copilot is using Bing search results to augment summaries.
D.Copilot is incorrectly configured to ignore meeting permissions.
AnswerB

Copilot uses data the user can access.

Why this answer

Option B is correct because Microsoft 365 Copilot generates meeting summaries by aggregating content from meeting artifacts such as transcripts, recordings, and shared calendars. If a user has access to a meeting's transcript or recording (e.g., via a shared calendar or because the meeting was recorded and stored in a location the user can access), Copilot can include that meeting's information in summaries even if the user did not attend. This behavior is by design, as Copilot respects existing permissions on the underlying data.

Exam trap

The trap here is that candidates often assume Copilot uses meeting attendance or organizer permissions to filter summaries, when in fact it relies on the underlying permissions of the meeting's artifacts (transcripts, recordings, calendar items), which can be broader than the attendee list.

How to eliminate wrong answers

Option A is wrong because granting everyone view access to a meeting would allow users to see the meeting details, but Copilot does not automatically include meetings in summaries based solely on view access; it requires access to the meeting's artifacts like transcripts or recordings. Option C is wrong because Copilot does not use Bing search results to augment meeting summaries; it relies on the user's Microsoft Graph data and permissions, not external web searches. Option D is wrong because there is no configuration setting in Copilot to 'ignore meeting permissions'; Copilot strictly adheres to the permissions set on the meeting artifacts and does not have a mode that bypasses them.

756
MCQmedium

Your organization is building a chatbot using Azure OpenAI Service. The chatbot must provide citations from a set of internal documents stored in Azure Blob Storage. You need to configure the solution to minimize token usage while ensuring citations are accurate. Which approach should you use?

A.Embed all document content into the system prompt
B.Fine-tune a model on the documents so it can recall them from memory
C.Use a large context window model (e.g., 32K) and include all documents in the prompt
D.Use Azure OpenAI on your data with Azure Cognitive Search for hybrid retrieval
AnswerD

Hybrid retrieval reduces token usage by fetching only relevant chunks.

Why this answer

Option B is correct because Azure OpenAI on your data with Azure Cognitive Search uses a hybrid retrieval approach (vector + keyword), which is more token-efficient than embedding all content into the prompt. Option A is wrong because embedding entire documents wastes tokens. Option C is wrong because it lacks retrieval.

Option D is wrong because fine-tuning does not support dynamic document citation.

757
MCQmedium

You are using Azure OpenAI to generate product descriptions. You notice that the descriptions are often too similar to each other. Which parameter should you adjust to increase diversity?

A.Increase the temperature value.
B.Decrease the top_p value.
C.Increase the max_tokens value.
D.Increase the frequency_penalty value.
AnswerA

Higher temperature increases randomness, leading to more diverse outputs.

Why this answer

Increasing the temperature parameter makes the model more creative by raising the probability of sampling lower-probability tokens, which increases diversity in the generated text. A higher temperature (e.g., 0.9) flattens the probability distribution, so the model is less likely to always pick the most probable next word, resulting in more varied outputs.

Exam trap

Microsoft often tests the distinction between temperature (which controls randomness/creativity) and frequency_penalty (which controls repetition), leading candidates to mistakenly choose frequency_penalty when the question asks for diversity in content rather than just avoiding repetition.

How to eliminate wrong answers

Option B is wrong because decreasing top_p (nucleus sampling) reduces the cumulative probability mass considered for token selection, which actually makes outputs less diverse by focusing only on the most likely tokens. Option C is wrong because increasing max_tokens only extends the maximum length of the generated response; it does not affect the randomness or diversity of token choices. Option D is wrong because increasing frequency_penalty reduces the likelihood of repeating the same tokens or phrases, which can increase lexical diversity but does not directly control the overall creativity or randomness of the output like temperature does.

758
Multi-Selecteasy

Which TWO of the following are valid ways to authenticate to Azure AI services?

Select 3 answers
A.Subscription key (API key)
B.Shared access signature (SAS) token
C.OAuth 2.0 client credentials flow with Microsoft Entra ID
D.Managed Identity with Microsoft Entra ID
E.X.509 certificate
AnswersA, C, D

API keys are a primary authentication method for Azure AI services.

Why this answer

Option A is correct because Azure AI services accept subscription keys (API keys) as a primary authentication method. When you create an Azure AI resource, the portal generates two keys that can be passed in the 'Ocp-Apim-Subscription-Key' header to authenticate requests. This is the simplest and most common method for initial development and testing.

Exam trap

The trap here is that candidates often confuse SAS tokens (which are for Azure Storage) with subscription keys (for AI services), and they may overlook that Managed Identity is a valid Entra ID authentication method, leading them to incorrectly select only one correct answer or pick SAS as a second option.

759
MCQmedium

You submit the request shown in the exhibit to the Azure AI Language service. What will the response contain?

A.Only extracted key phrases.
B.Entity recognition results and sentiment scores.
C.Both key phrases and sentiment scores.
D.Only sentiment scores.
AnswerC

Both tasks are included.

Why this answer

Option C is correct because the request includes two tasks: KeyPhraseExtraction and SentimentAnalysis, so the response will contain key phrases and sentiment scores. Option A is incorrect because it only mentions key phrases. Option B is incorrect because it only mentions sentiment.

Option D is incorrect because it mentions entities which are not requested.

760
Multi-Selecthard

You are building a document processing solution that extracts information from invoices. The invoices come in various formats and languages. You need to extract line items, totals, and supplier names. Which THREE services should you combine?

Select 3 answers
A.Azure AI Custom Vision
B.Azure AI Translator
C.Azure AI Content Safety
D.Azure AI Document Intelligence
E.Azure AI Vision OCR
AnswersB, D, E

Translates text if invoices are in multiple languages.

Why this answer

Document Intelligence extracts structured data. AI Vision OCR handles images. AI Translator translates if needed.

These three work together.

761
MCQeasy

Refer to the exhibit. An Azure Cognitive Services Computer Vision API call for image captioning is returning only one caption. The developer wants to get three possible captions ranked by confidence. Which parameter should be modified in the request?

A.Use a different API version, such as 2023-04-01.
B.Modify the URL to point to a different image.
C.Change the language parameter to 'multi'.
D.Set the maxCandidates value to 3.
AnswerD

maxCandidates defines how many captions the API returns.

Why this answer

The `maxCandidates` parameter in the Computer Vision Image Analysis API controls the maximum number of captions returned in the response. By default, this value is 1, so only the top-ranked caption is returned. Setting `maxCandidates=3` instructs the API to return up to three captions, each with its own confidence score, ranked from highest to lowest confidence.

Exam trap

The trap here is that candidates may confuse the `maxCandidates` parameter with other parameters like `language` or `details`, or assume that changing the API version or image source would increase the number of captions, when in fact the default behavior is to return only one caption unless explicitly overridden.

How to eliminate wrong answers

Option A is wrong because changing the API version (e.g., to 2023-04-01) does not affect the number of captions returned; the `maxCandidates` parameter is available across supported versions. Option B is wrong because pointing to a different image changes the input but does not alter the request parameter that controls the number of captions; the API would still return only one caption per image unless `maxCandidates` is set. Option C is wrong because the `language` parameter specifies the language of the returned text (e.g., 'en' for English), not the count of captions; 'multi' is not a valid language value for this API.

762
MCQmedium

A company builds a knowledge mining solution using Azure AI Search with a custom skillset that includes an OCR skill. They want to ensure that images embedded in PDFs are processed. What should they configure?

A.Set the 'defaultLanguageCode' to 'en'
B.Set the 'textExtractionAlgorithm' to 'printed'
C.Set the 'imageAction' parameter to 'generateNormalizedImages'
D.Set the 'lineEnding' parameter to 'space'
AnswerC

This parameter enables extraction of images from documents.

Why this answer

Option C is correct because the 'imageAction' parameter in Azure AI Search's OCR skill controls whether images embedded in documents (including PDFs) are extracted and processed. Setting it to 'generateNormalizedImages' ensures that images within PDFs are normalized and passed to the OCR skill for text extraction, which is essential for processing embedded images.

Exam trap

The trap here is that candidates may confuse parameters that affect OCR output formatting (like 'lineEnding' or 'defaultLanguageCode') with the parameter that actually enables image extraction from PDFs, leading them to overlook the 'imageAction' setting.

How to eliminate wrong answers

Option A is wrong because 'defaultLanguageCode' specifies the language for text recognition, not whether images are extracted from PDFs; it does not enable image processing. Option B is wrong because 'textExtractionAlgorithm' determines the OCR algorithm (e.g., 'printed' or 'handwritten') but does not control the extraction of images from PDFs; it only affects how text is recognized once images are available. Option D is wrong because 'lineEnding' parameter controls the line break character in OCR output (e.g., 'space', 'carriageReturn'), which is irrelevant to enabling image extraction from PDFs.

763
MCQmedium

Refer to the exhibit. An administrator runs this Azure CLI command to deploy a GPT-4 model in Azure AI Foundry. The command fails with an error that the deployment name already exists. What should the administrator do to resolve the issue?

A.Use a different deployment name or delete the existing deployment.
B.Specify a different resource group.
C.Remove the --sku-name parameter.
D.Use a different model version.
AnswerA

Deployment names must be unique within an Azure AI Foundry resource.

Why this answer

The error message indicates that a deployment with the same name already exists in the Azure AI Foundry workspace. In Azure AI Foundry, deployment names must be unique within a workspace. The correct resolution is to either choose a different deployment name or delete the existing deployment before re-running the command.

This aligns with the Azure CLI behavior where resource names (including AI model deployments) must be unique per scope.

Exam trap

The trap here is that candidates may think the error is about model availability or SKU constraints, but the error explicitly states 'deployment name already exists,' which is a naming conflict, not a capacity or version issue.

How to eliminate wrong answers

Option B is wrong because specifying a different resource group does not resolve a deployment name conflict within the same workspace; the deployment name uniqueness is scoped to the workspace, not the resource group. Option C is wrong because removing the --sku-name parameter would change the pricing tier or capacity, but does not address the duplicate name error. Option D is wrong because using a different model version does not change the deployment name; the conflict is on the name, not the model version.

764
Multi-Selectmedium

Which THREE components are required to implement a Retrieval-Augmented Generation (RAG) solution with Azure OpenAI Service? (Choose three.)

Select 3 answers
A.An embedding model (e.g., text-embedding-ada-002)
B.A fine-tuned model
C.An Azure OpenAI Service model (LLM)
D.Azure AI Content Safety
E.A vector database (e.g., Azure AI Search)
AnswersA, C, E

Embedding models convert documents into vector representations.

Why this answer

Option A is correct because the LLM is the core model. Option C is correct because an embedding model converts documents to vectors. Option E is correct because a vector database stores and retrieves embeddings.

Option B is wrong as Azure AI Content Safety is optional. Option D is wrong as fine-tuning is not required for RAG.

765
MCQeasy

You need to create a chatbot that uses Azure OpenAI to answer questions about your company's internal policies. The responses must be based only on the provided policy documents. Which approach should you use?

A.Use the model's pre-existing knowledge about common policies.
B.Fine-tune a GPT model on the policy documents.
C.Use prompt engineering to instruct the model to only use policy knowledge.
D.Use Retrieval-Augmented Generation (RAG) with an Azure AI Search index of the documents.
AnswerD

RAG ensures responses are grounded in the retrieved documents.

Why this answer

Option B is correct because RAG uses provided documents as source. Option A is wrong because fine-tuning on policies may still generate ungrounded responses. Option C is wrong because prompt engineering alone does not enforce grounding.

Option D is wrong because the model's training data is generic.

766
Multi-Selecthard

You are developing a custom text classification model using Azure AI Language. You have labeled 2000 documents across 10 categories. You need to evaluate the model's performance before deploying to production. Which THREE metrics should you examine?

Select 3 answers
A.Recall
B.Word Error Rate
C.BLEU Score
D.F1 Score
E.Precision
AnswersA, D, E

Measures the proportion of actual positives correctly identified.

Why this answer

Options A, B, and D are correct because precision, recall, and F1 score are the standard metrics for classification models. Option C is wrong because BLEU is for machine translation. Option E is wrong because Word Error Rate is for speech recognition.

767
MCQmedium

You are using Azure OpenAI Service with the system message shown in the exhibit. The model sometimes answers questions using general knowledge even when the context does not contain the answer. Which modification should you make to enforce the behavior?

A.Add a user message repeating the instruction
B.Reduce the temperature to 0
C.Fine-tune the model with examples of refusing to answer
D.Set the 'strict' parameter to true in Azure OpenAI On Your Data configuration
AnswerD

The strict setting forces the model to use only the provided context.

Why this answer

Option D is correct because the 'strict' parameter in Azure OpenAI On Your Data configuration forces the model to rely exclusively on the provided data sources and refuse to answer when the context lacks the information. This directly enforces the desired behavior of not falling back on general knowledge.

Exam trap

Microsoft often tests the misconception that prompt engineering (like repeating instructions) or temperature adjustments can enforce strict data-only behavior, when in reality the 'strict' parameter in Azure OpenAI On Your Data is the specific mechanism designed for this purpose.

How to eliminate wrong answers

Option A is wrong because adding a user message repeating the instruction does not override the model's inherent tendency to use general knowledge; it only provides a prompt-level hint that can be ignored. Option B is wrong because reducing the temperature to 0 makes the model more deterministic but does not prevent it from generating answers from its pre-trained knowledge when the context is insufficient. Option C is wrong because fine-tuning with examples of refusing to answer requires custom training data and is not a direct configuration for Azure OpenAI On Your Data; it also does not enforce strict context-only behavior at inference time.

768
MCQmedium

You are a developer at a legal firm. The firm has a repository of court case documents stored as PDFs in Azure Blob Storage. You need to build a knowledge mining solution that enables lawyers to search for cases by parties involved, judge name, case number, date, and key legal topics. The documents are in English and contain both typed and handwritten text. The solution must extract the aforementioned metadata and also identify citations to other cases (e.g., 'Smith v. Jones'). You plan to use Azure AI Search with cognitive skills. Which combination of skills should you include in your skillset?

A.OCR skill (to handle handwriting), Key Phrase Extraction skill, Sentiment skill, and Language Detection skill.
B.OCR skill, Document Layout skill, Entity Recognition skill, and Text Translation skill.
C.Document Layout skill, Entity Recognition skill (for person names and organizations), Custom Entity Lookup skill (for case citation patterns), and Key Phrase Extraction skill for legal topics.
D.Document Layout skill, Text Translation skill, Sentiment skill, and Key Phrase Extraction skill.
AnswerC

Covers all requirements: extraction of metadata, citations, and topics.

Why this answer

Option A is correct because Document Layout skill handles typed and handwritten text, Entity Recognition can extract person names (parties, judge) and law-related entities, and a custom skill can parse case citations. Option B misses entity extraction for parties and judge. Option C uses OCR but Document Layout is better.

Option D uses Translator unnecessarily.

769
MCQeasy

You are using Azure AI Language to perform sentiment analysis on customer feedback. You need to monitor the number of requests sent to the endpoint per minute. Which Azure service should you use?

A.Azure Monitor
B.Azure Cost Management
C.Azure Advisor
D.Azure Log Analytics
AnswerA

Azure Monitor collects and analyzes metrics like requests per minute.

Why this answer

Azure Monitor is the correct service because it provides platform metrics, including the 'Requests Per Minute' metric for Azure AI Language endpoints. This metric is automatically collected and can be viewed in the Azure portal, set up alerts on, or queried via the Azure Monitor REST API. It directly answers the need to monitor request rate without additional configuration.

Exam trap

The trap here is that candidates often confuse Azure Monitor with Azure Log Analytics, assuming that all monitoring requires log ingestion, when in fact Azure Monitor provides built-in metrics without the need for logging or additional costs.

How to eliminate wrong answers

Option B (Azure Cost Management) is wrong because it focuses on tracking and optimizing cloud spending, not on real-time operational metrics like request rate. Option C (Azure Advisor) is wrong because it provides best-practice recommendations for cost, security, reliability, and performance, but does not expose per-minute request counts. Option D (Azure Log Analytics) is wrong because while it can ingest and query logs, it is not the primary service for out-of-the-box metrics; Azure Monitor collects metrics natively, and Log Analytics is used for log-based data, not for simple metric monitoring like requests per minute.

770
MCQmedium

A company uses Azure AI Speech to transcribe customer service calls. They need to identify callers by name and account number. Which feature should be enabled?

A.Speaker recognition
B.Pronounciation assessment
C.Batch transcription
D.Custom speech model
AnswerD

Custom speech models improve recognition of domain-specific terms.

Why this answer

Option A is correct because custom speech models improve recognition of domain-specific terms like names and account numbers. Option B is wrong because speaker recognition identifies who is speaking, not specific words. Option C is wrong because pronounciation assessment evaluates pronunciation, not recognition.

Option D is wrong because batch transcription is for large volumes, not real-time.

771
MCQeasy

You are deploying a chatbot using Azure AI Bot Service and Language Understanding (LUIS). The bot must understand user intent from free-text input. Which component should you train?

A.Language Understanding (LUIS) model
B.Speech-to-text model
C.QnA Maker knowledge base
D.Computer Vision model
AnswerA

LUIS extracts intents and entities from free-text input.

Why this answer

The Language Understanding (LUIS) model is the correct component to train because the bot needs to interpret free-text user input and extract intent. LUIS is a natural language processing service specifically designed for intent recognition and entity extraction from conversational phrases. Training the LUIS model with labeled utterances teaches it to map user expressions to predefined intents, enabling the bot to understand and respond appropriately.

Exam trap

The trap here is that candidates may confuse the role of LUIS with QnA Maker, assuming both handle any text input, but LUIS is for intent classification from free-text conversation, while QnA Maker is for retrieving answers from a fixed knowledge base, not for understanding dynamic user intents.

How to eliminate wrong answers

Option B is wrong because a Speech-to-text model converts audio to text, but the question specifies free-text input, not spoken input; training this model would be unnecessary and irrelevant for text-based intent understanding. Option C is wrong because QnA Maker knowledge base is designed for answering factual questions from a structured FAQ or document, not for understanding free-form intents from conversational input; it lacks the intent classification capability required here. Option D is wrong because a Computer Vision model processes images and video, not text; it has no role in interpreting user intent from free-text input.

772
Multi-Selectmedium

You are building a solution that must translate customer chat messages from Spanish to English in real-time. The solution must also detect the language of incoming messages to handle cases where users write in other languages. Which TWO Azure AI service features should you use?

Select 2 answers
A.Azure AI Translator - Language Detection
B.Azure AI Translator - Translation
C.Azure AI Speech - Speech Translation
D.Azure AI Language Understanding (LUIS)
E.Azure AI Custom Question Answering
AnswersA, B

Detects the language of text input.

Why this answer

Options A and D are correct because the Azure AI Translator service provides both translation and language detection. Option B is wrong because the Speech service is for audio, not text. Option C is wrong because the LUIS is for intent recognition, not translation.

Option E is wrong because the QnA Maker is for question answering, not translation.

773
MCQeasy

A developer is building a mobile app that uses Azure Computer Vision to analyze images. The app needs to handle many requests with low latency. Which pricing tier should they choose?

A.Video Analyzer S1 tier.
B.Free F0 tier.
C.Custom Vision S0 tier.
D.Computer Vision S1 tier.
AnswerD

S1 offers up to 30 calls per second.

Why this answer

The Computer Vision S1 tier is designed for production workloads requiring high throughput and low latency, making it suitable for a mobile app that handles many image analysis requests. Unlike the Free F0 tier, which has strict rate limits (e.g., 20 transactions per minute), the S1 tier offers higher transactions per second (e.g., up to 10 TPS) and guaranteed performance for real-time scenarios.

Exam trap

The trap here is that candidates may confuse the Computer Vision S1 tier with the Custom Vision S0 tier or the Video Analyzer S1 tier, not realizing that each Azure AI service has its own distinct pricing tiers and that the question specifically targets the Computer Vision service for image analysis.

How to eliminate wrong answers

Option A is wrong because Video Analyzer S1 is a pricing tier for Azure Video Analyzer, a service for video indexing and analysis, not for image analysis with Computer Vision. Option B is wrong because the Free F0 tier has severe rate limits (e.g., 20 calls per minute) and is intended only for evaluation and small-scale testing, not for production apps with many requests. Option C is wrong because Custom Vision S0 is a tier for the Custom Vision service, which is used to train and deploy custom image classification models, not the general-purpose Computer Vision API for analyzing images.

774
MCQmedium

You are designing an Azure AI solution that uses Language Understanding (LUIS) for intent detection. The solution must handle multiple languages dynamically based on the user's locale. What should you do?

A.Use Azure Translator to translate user input to English before sending to LUIS.
B.Create separate LUIS applications for each language and route based on locale.
C.Train a single LUIS app with utterances in all languages.
D.Enable the 'Multi-Language' feature in the LUIS app.
AnswerB

This is the recommended approach for multi-language support.

Why this answer

Option B is correct because LUIS does not natively support multi-language within a single application; each LUIS app is designed for a single language. To handle multiple languages dynamically, you must create separate LUIS applications for each language and route user utterances based on the detected locale, ensuring accurate intent and entity recognition per language.

Exam trap

The trap here is that candidates assume LUIS has a built-in multi-language feature or that translation is a viable shortcut, but Microsoft explicitly requires separate LUIS apps per language and does not support multi-language training within a single app.

How to eliminate wrong answers

Option A is wrong because translating user input to English before sending to LUIS introduces translation latency, potential loss of nuance, and inaccuracies in intent detection, as LUIS is optimized for native language patterns. Option C is wrong because training a single LUIS app with utterances in multiple languages degrades performance, as LUIS expects consistent language structure and cannot distinguish between languages during prediction. Option D is wrong because there is no 'Multi-Language' feature in LUIS; the platform requires separate apps for each language, and enabling such a feature would not resolve the fundamental single-language limitation.

775
Multi-Selecthard

A company is planning to use Azure AI Document Intelligence to extract data from invoices. The solution must handle variations in invoice formats. Which TWO actions should be taken to improve accuracy?

Select 2 answers
A.Avoid using custom models; rely on prebuilt models only.
B.Include samples of different layouts in the training set.
C.Use a large number of unlabeled samples.
D.Use a single prebuilt model for all invoices.
E.Train a custom model with labeled invoice samples.
AnswersB, E

Diverse layouts improve generalization.

Why this answer

Option B is correct because including samples of different invoice layouts in the training set enables the custom model to learn variations in structure, field placement, and formatting. This improves the model's ability to generalize across diverse invoice formats, reducing extraction errors. Azure AI Document Intelligence custom models require labeled training data to adapt to specific document types, and diverse samples directly address format variability.

Exam trap

The trap here is that candidates often assume prebuilt models are sufficient for all invoice formats, ignoring that custom models with diverse labeled samples are necessary to handle layout variations and achieve high accuracy.

776
Multi-Selecthard

Which TWO of the following are best practices for managing Azure AI services costs?

Select 2 answers
A.Use the S0 pricing tier for production workloads
B.Always use the Free tier to avoid charges
C.Scale up partitions to improve performance
D.Increase batch size to reduce number of API calls
E.Set up budget alerts in Azure Cost Management
AnswersA, E

S0 provides predictable pricing for production.

Why this answer

Option A is correct because the S0 pricing tier is the standard paid tier for Azure AI services, such as Cognitive Services, and is designed for production workloads. It provides higher throughput, reliability, and support compared to lower tiers, making it a cost-effective choice for production scenarios where performance and availability are critical.

Exam trap

The trap here is that candidates often confuse cost-saving strategies (like using the Free tier or batching) with best practices for managing costs in production, overlooking that the Free tier is not for production and that batching may not be applicable or effective for all services.

777
MCQhard

You are planning an Azure AI solution that uses Azure AI Document Intelligence to extract data from scanned PDFs. The solution must support both structured and unstructured documents. The extracted data must be validated against a predefined schema before being stored in Azure Cosmos DB. What should you use for schema validation?

A.Azure Functions to parse and validate the extracted data.
B.Azure AI Document Intelligence custom extraction model with a defined schema.
C.Azure AI Document Intelligence prebuilt model for invoices.
D.Azure Logic Apps with a JSON schema validation step.
AnswerB

Custom extraction models allow defining a schema and validating extracted data.

Why this answer

Option B is correct because Azure AI Document Intelligence custom extraction models allow you to define a schema (field names, types, and relationships) that the service uses to extract and validate data from both structured and unstructured documents. This built-in schema validation ensures extracted data conforms to your predefined structure before it is stored in Azure Cosmos DB, eliminating the need for additional validation logic.

Exam trap

The trap here is that candidates often assume prebuilt models (like invoices) are sufficient for schema validation, but they lack the ability to enforce a custom schema across both structured and unstructured documents, which is the core requirement in this question.

How to eliminate wrong answers

Option A is wrong because Azure Functions would require custom code to parse and validate extracted data, adding complexity and maintenance overhead, whereas the schema validation should be handled natively by the extraction service. Option C is wrong because the prebuilt invoice model is designed only for structured invoice documents and cannot handle unstructured documents or support a custom predefined schema for validation. Option D is wrong because Azure Logic Apps with a JSON schema validation step validates the structure of the JSON payload after extraction, but does not validate the extracted data against the original document's schema or enforce business rules at the extraction point.

778
MCQmedium

You are implementing an agentic solution using Azure AI Agent Service. The agent needs to maintain conversation context across multiple turns. You configure the agent with a custom prompt that includes a 'system message' and 'few-shot examples'. However, after a few turns, the agent starts repeating the same responses. What is the most likely cause?

A.The 'max_tokens' parameter is set too low for the response
B.The 'max_context_length' is set too low, causing earlier turns to be truncated
C.The agent is using a 'context recycling' feature that resets after each turn
D.The temperature is set too high, causing the model to become deterministic
AnswerB

Truncation loses context, leading to repetition.

Why this answer

Option B is correct because a low maximum context length truncates earlier conversation turns, causing the agent to lose context and repeat responses. Option A is incorrect because the token limit per request affects the length of individual responses, not repetition. Option C is incorrect because the Azure OpenAI API does not recycle context unless explicitly configured.

Option D is incorrect because changing temperature affects randomness, not repetition due to context loss.

779
MCQmedium

Refer to the exhibit. You send this request to the Azure AI Language Service for custom entity recognition. The response returns no entities. What is the most likely reason?

A.The text input is too short for entity recognition
B.The language parameter is set incorrectly
C.The model version is not specified
D.The projectName and deploymentName are missing from the body parameters
AnswerD

Custom entity recognition requires project and deployment names in the body.

Why this answer

The correct answer is D because the request is missing the 'projectName' and 'deploymentName' under the 'parameters' of the body. The exhibit shows them at the top level but not inside the body parameters. The API requires these inside the body.

A (language mismatch) is not the issue because 'en' is correct. B (text too short) is not the cause. C (model version) is set to 'latest'.

780
MCQhard

Your company has an existing Azure AI Search index used for e-commerce product search. The index is populated by an indexer that runs every hour. Recently, the product team added a new field 'promotionPrice' that should be searchable. You need to update the index to include this field without reindexing all documents. The field should be of type Edm.Double, searchable, filterable, and sortable. What should you do?

A.Add the new field to the data source mapping in the indexer. The indexer will automatically update the index schema.
B.Update the index schema to add the new field using the portal or REST API. Then update the indexer to map the field.
C.Change the indexer schedule to run every 5 minutes to pickup the new field faster.
D.Delete the index and recreate it with the new field. Then run the indexer to repopulate all documents.
AnswerB

No downtime; incremental update.

Why this answer

Option B is correct because Azure AI Search allows you to add new fields to an existing index schema without reindexing all documents, as long as the new field is nullable or has a default value. After updating the index schema via the portal or REST API, you must update the indexer's field mappings to map the new 'promotionPrice' field from the data source to the index. The indexer will then populate the field for existing documents on its next run, and new documents will include the field automatically.

Exam trap

The trap here is that candidates assume the indexer can automatically update the index schema when a new field is added to the data source, but Azure AI Search requires explicit schema updates before the indexer can map and populate the new field.

How to eliminate wrong answers

Option A is wrong because the indexer's data source mapping only defines how source fields map to index fields; it cannot modify the index schema itself. Option C is wrong because changing the indexer schedule does not add new fields to the index; the indexer can only write to fields that already exist in the index schema. Option D is wrong because deleting and recreating the index is unnecessary and causes downtime; Azure AI Search supports adding new fields to an existing index without full reindexing, provided the new field is nullable or has a default value.

781
Multi-Selectmedium

Which TWO configurations are required to enable Azure AI Search to index content from an Azure SQL database?

Select 2 answers
A.Create a custom skillset for data enrichment
B.Configure semantic ranking on the index
C.Enable change tracking on the Azure SQL table
D.Define a data source connection to the Azure SQL database
E.Enable high availability on the Azure SQL database
AnswersC, D

Change tracking allows the indexer to detect new, updated, or deleted rows for incremental indexing.

Why this answer

Options A and C are correct. A: A data source connection is needed to define the source. C: A skillset is optional but required if enrichment is needed; however, the question asks for required configurations, but a skillset is not required for basic indexing.

Actually, for basic indexing, only data source and index with indexer are required. However, the question may imply enrichment. But typical required: data source, index, indexer.

Skillset optional. So A and C? Wait, skillset is not required. Let's re-evaluate: For basic indexing, you need a data source definition and an index.

The indexer ties them together. Skillset is optional. So required: data source and index.

But options: A (data source), C (skillset) - skillset not required. D (index) is not listed. Options: A: data source, B: high availability, C: skillset, D: semantic configuration, E: change tracking.

Required for SQL indexing: data source (A) and change tracking (E) to detect new/modified rows. Skillset not required. So correct: A and E.

Yes.

782
MCQmedium

Your organization is using Azure OpenAI Service to generate content. You need to ensure that the content meets safety guidelines by filtering harmful outputs. What should you configure?

A.Enable the Responsible AI dashboard.
B.Configure the content filters in the Azure OpenAI Studio.
C.Use Azure AI Content Safety APIs to analyze outputs.
D.Set the system message to instruct the model to avoid harmful content.
AnswerB

Azure OpenAI provides configurable content filters.

Why this answer

Option B is correct because content filters in Azure OpenAI Studio allow you to define severity levels (safe, low, medium, high) for categories like hate, sexual, violence, and self-harm, which are enforced at the inference API level to block or flag harmful outputs before they reach the user. This is the primary configuration for filtering model-generated content in Azure OpenAI Service.

Exam trap

The trap here is that candidates often confuse the Responsible AI dashboard (a monitoring tool) with active content filtering, or they assume that system messages alone are sufficient for safety, when in fact content filters provide the only guaranteed enforcement layer at the API level.

How to eliminate wrong answers

Option A is wrong because the Responsible AI dashboard is a monitoring and reporting tool that provides visibility into model behavior and fairness metrics, but it does not actively filter or block harmful outputs in real-time. Option C is wrong because Azure AI Content Safety APIs are a separate service for analyzing user-generated or third-party content, not for filtering outputs from Azure OpenAI models directly; they would require an additional integration layer. Option D is wrong because system messages are instructional prompts that guide model behavior but are not a reliable enforcement mechanism—they can be overridden by adversarial inputs or model quirks, and they lack the deterministic filtering capabilities of content filters.

783
MCQmedium

A developer uses the Azure OpenAI SDK to generate code snippets. The generated code sometimes contains security vulnerabilities. What is the most effective way to mitigate this risk?

A.Set the temperature parameter to 0 to make the output deterministic.
B.Post-process the generated code using a static code analysis tool.
C.Fine-tune the model on a dataset of secure code examples.
D.Include a system message that instructs the model to avoid insecure coding patterns.
AnswerD

System messages can guide the model to produce safer code.

Why this answer

Option D is correct because a system message sets the behavioral context for the model at inference time, instructing it to avoid insecure coding patterns without requiring retraining. This is the most direct and effective mitigation as it leverages the model's instruction-following capability to reduce vulnerabilities in generated code, aligning with Azure OpenAI's content filtering and safety system guidance.

Exam trap

Microsoft often tests the misconception that fine-tuning is the only way to customize model behavior, but the trap here is that a system message is a simpler, more flexible, and equally effective method for guiding output without the overhead of retraining.

How to eliminate wrong answers

Option A is wrong because setting temperature to 0 makes the output deterministic but does not address security; it only reduces randomness, not insecure patterns. Option B is wrong because post-processing with static analysis detects vulnerabilities after generation but does not prevent them at the source, adding latency and requiring separate tooling. Option C is wrong because fine-tuning on secure code examples is resource-intensive, requires a curated dataset, and may not generalize to all insecure patterns; it is less practical than a simple system message for immediate risk mitigation.

784
MCQmedium

You need to create a solution that extracts key-value pairs from scanned invoices using Azure AI Document Intelligence. The invoices have varying layouts. Which model should you use?

A.Layout model
B.Custom extraction model
C.Read model
D.Prebuilt invoice model
AnswerD

Prebuilt invoice model is trained on many invoice layouts and extracts key-value pairs.

Why this answer

The Prebuilt invoice model (Option D) is specifically designed to extract key-value pairs, line items, and other structured fields from invoices, even when layouts vary. It is trained on thousands of invoice samples and uses deep learning to handle diverse formats without requiring custom training, making it the correct choice for this scenario.

Exam trap

The trap here is that candidates often confuse the Layout model's ability to extract tables and text with the specific key-value pair extraction needed for invoices, leading them to choose Option A instead of the purpose-built Prebuilt invoice model.

How to eliminate wrong answers

Option A is wrong because the Layout model extracts text, tables, and selection marks but does not extract key-value pairs or invoice-specific fields like invoice number or vendor details. Option B is wrong because a Custom extraction model requires labeled training data and is overkill when a prebuilt model already exists for invoices; it is intended for documents not covered by prebuilt models. Option C is wrong because the Read model only extracts printed and handwritten text (OCR) without any structure or key-value pair extraction.

785
MCQmedium

A retail company uses Azure Computer Vision to analyze customer traffic in stores. They deploy a custom object detection model to count customers and detect occupancy. After deployment, the model consistently underestimates the number of customers during peak hours. The company has retrained the model with more data but the issue persists. What is the most likely cause?

A.The model is not being batch-processed for inference.
B.The training data does not adequately represent peak-hour scenarios.
C.The model is overfitting to the training data.
D.The Computer Vision API version is outdated.
AnswerB

Data drift or lack of representative samples for peak hours leads to underestimation during those times.

Why this answer

The model consistently underestimates customer counts during peak hours, which indicates a distribution shift between the training data and the inference environment. Even after retraining with more data, the issue persists because the additional data likely still lacks sufficient representation of peak-hour scenarios (e.g., high density, occlusion, rapid movement). In Azure Custom Vision, object detection models learn from labeled examples; if the training set does not include diverse peak-hour images with varied lighting, crowd densities, and angles, the model will fail to generalize to those conditions.

Exam trap

The trap here is that candidates may assume retraining with 'more data' automatically fixes the issue, but the key is that the additional data must be representative of the specific failure scenario (peak hours), not just any data.

How to eliminate wrong answers

Option A is wrong because batch processing affects throughput and latency, not the accuracy of individual inference results; the model's underestimation is a precision/recall issue, not a processing mode issue. Option C is wrong because overfitting would cause the model to perform well on training data but poorly on new data in general, not specifically during peak hours; the consistent underestimation only in peak hours points to a data distribution mismatch, not overfitting. Option D is wrong because the Computer Vision API version affects available features and endpoints, not the learned weights of a custom object detection model; the model's behavior is determined by its training data and architecture, not the API version used for deployment.

786
MCQmedium

Your team is developing a chatbot using Azure AI Bot Service with language understanding via Azure AI Language (CLU). You need to ensure that the chatbot can handle multiple intents in a single user utterance and return confidence scores for each. Which configuration should you use?

A.Use a Conversational Language Understanding (CLU) project with the 'Multiple intents' setting enabled.
B.Use an Orchestration workflow project that connects to multiple CLU projects.
C.Use a standard CLU project and manually combine intents.
D.Use Azure AI QnA Maker with custom logic to detect intents.
AnswerA

CLU supports multiple intents per utterance with confidence scores.

Why this answer

Option A is correct because Azure AI Language's Conversational Language Understanding (CLU) supports a 'Multiple intents' setting that allows the model to predict multiple intents from a single utterance, each with its own confidence score. This is the native way to handle multi-intent scenarios without requiring orchestration or custom logic.

Exam trap

The trap here is that candidates often confuse Orchestration workflow (which routes to different projects) with the ability to handle multiple intents within a single CLU project, leading them to choose Option B incorrectly.

How to eliminate wrong answers

Option B is wrong because an Orchestration workflow project routes utterances to different CLU projects or other services, but it does not enable multiple intents within a single CLU project; it selects one project per utterance. Option C is wrong because a standard CLU project without the 'Multiple intents' setting enabled will only return the top intent, not multiple intents with confidence scores; manually combining intents is not supported by the service. Option D is wrong because Azure AI QnA Maker is designed for FAQ-style question answering, not intent detection, and it lacks native multi-intent support; custom logic would be brittle and not leverage CLU's built-in capabilities.

787
MCQeasy

You are building a generative AI solution using Azure Machine Learning prompt flow. The solution must allow business analysts without coding experience to modify prompts and evaluate different model versions. What should you do?

A.Provide the analysts with a Jupyter notebook using the OpenAI Python SDK
B.Deploy a chatbot in Microsoft Copilot Studio and let analysts configure it
C.Implement a custom web UI using Azure Static Web Apps and Azure Functions
D.Use Azure Machine Learning prompt flow with the visual designer and variant management
AnswerD

Prompt flow's visual interface allows no-code prompt engineering and evaluation.

Why this answer

Option C is correct because prompt flow in Azure ML provides a visual designer and evaluation tools suitable for non-coders. Option A is wrong because direct API calls require coding. Option B is wrong because it's a different service.

Option D is wrong because it's not a visual tool.

788
MCQhard

Refer to the exhibit. You are using the Azure AI Face API to detect faces in an image. You need to ensure that the response includes the unique face ID for each detected face. However, the response does not contain face IDs. What is the most likely cause?

A.The 'returnFaceLandmarks' parameter must be set to true.
B.The 'returnFaceId' parameter is set to false.
C.The 'detectionModel' parameter is not set; the default detection model does not return face IDs.
D.The API version is incorrect; use '2023-06-01-preview' instead.
AnswerC

Detection model must be set to 'detection_03' for face IDs.

Why this answer

Option A is correct because the Face API requires that the 'detectionModel' parameter be set to 'detection_03' to return face IDs. The exhibit does not specify detectionModel, so it defaults to 'detection_01' which does not return face IDs. Option B is wrong because 'returnFaceId' is set to true.

Option C is wrong because the API version is valid. Option D is wrong because face landmarks are not required for face ID.

789
MCQeasy

A developer is creating a custom text classification model using Azure AI Language. The dataset has 10,000 documents across 50 categories. Which method is most suitable for labeling the data efficiently?

A.Use a prebuilt model from the Azure AI Language service
B.Use active learning in the custom text classification project
C.Manually label all documents in the Language Studio
D.Use Azure Machine Learning designer to auto-label
AnswerB

Active learning suggests labels, reducing manual effort.

Why this answer

Option B is correct because Azure AI Language supports active learning, which suggests labels and reduces manual effort. Option A is wrong because manual labeling of 10k docs is inefficient. Option C is wrong because Azure Machine Learning designer is not designed for text classification labeling.

Option D is wrong because prebuilt models cannot handle custom categories.

790
MCQhard

You are a developer at a large retail company. The company receives thousands of product reviews daily. You need to build a solution that automatically categorizes reviews into positive, negative, and neutral sentiments, and also extracts key product features mentioned (e.g., battery life, screen quality) along with their associated sentiments. The solution must be scalable and cost-effective. You have access to Azure AI Language. You decide to use the built-in sentiment analysis and opinion mining features. However, after initial testing, you find that the opinion mining feature does not always correctly associate sentiments with the correct product features. For example, in the review 'The battery life is great but the screen is terrible', opinion mining might incorrectly associate 'terrible' with 'battery life'. You need to improve the accuracy of feature-sentiment association. What should you do?

A.Create a custom NER project in Azure AI Language to extract product features, then use the opinion mining results and post-process to associate sentiments with the extracted features.
B.Use the PII recognition feature to identify product features as entities.
C.Use Conversational Language Understanding (CLU) to define intents for each product feature and train a model with labeled utterances.
D.Use the standard sentiment analysis API without opinion mining, and then use key phrase extraction to identify features and assign overall sentiment.
AnswerA

Custom NER can accurately extract the product features, and you can then use opinion mining scores to assign sentiment to each feature.

Why this answer

Option B is correct because by training a custom NER model to extract product features, you can then use the opinion mining results but map them to your custom entities more accurately. Option A is wrong because switching to CLU is for intent classification, not for fine-grained aspect sentiment. Option C is wrong because sentiment analysis without opinion mining loses aspect-level detail.

Option D is wrong because PII detection is irrelevant.

791
MCQmedium

A healthcare provider uses Azure Computer Vision to analyze medical images. They need to ensure patient data is not stored outside the Azure region. What should you configure?

A.Use the Free tier for Computer Vision.
B.Enable customer-managed keys (CMK) for the Computer Vision resource.
C.Deploy Computer Vision in multiple regions.
D.Configure a private endpoint for the Computer Vision resource.
AnswerD

Ensures data stays within the virtual network and region.

Why this answer

Option D is correct because a private endpoint ensures data stays within the virtual network and region. Option A is wrong because customer-managed keys encrypt data at rest but don't restrict storage location. Option B is wrong because the Free tier has usage limits and no regional guarantee.

Option C is wrong because multi-region deployment replicates data, opposite of requirement.

792
Multi-Selecthard

Which THREE factors should you consider when designing a knowledge mining solution that uses Azure AI Search and custom skills to extract insights from large volumes of documents?

Select 3 answers
A.The number of knowledge store projections affects indexing speed
B.The maximum execution time of the custom skill must fit within the indexer timeout
C.Incremental enrichment should be enabled to avoid reprocessing unchanged documents
D.Semantic ranking configuration must be included in the skillset
E.The custom skill should be stateless and idempotent to allow parallel execution
AnswersB, C, E

Long-running skills can cause timeouts and failures.

Why this answer

Options A, C, and E are correct. A: Skillset execution time must be within indexer timeout limits. C: Azure Functions can scale to handle concurrent requests.

E: Incremental enrichment reduces reprocessing. Option B is wrong because knowledge stores are for structured output, not performance. Option D is wrong because semantic ranking is for search relevance, not extraction.

793
MCQhard

Your organization is deploying a generative AI solution using Azure AI Foundry. The solution must comply with responsible AI principles, including fairness and transparency. Which combination of tools should you use to assess and mitigate bias in the model?

A.Azure AI Content Safety and Azure Machine Learning fairness assessment
B.Azure AI Language PII detection and Azure AI Search
C.Microsoft Defender XDR and Azure AI Language
D.Microsoft Purview Data Map and Azure AI Document Intelligence
AnswerA

Content Safety filters harmful content and fairness assessment evaluates bias.

Why this answer

Option B is correct because Azure AI Content Safety provides content filtering, and the fairness assessment in Azure Machine Learning evaluates bias. Option A is wrong because Microsoft Purview is for data governance, not bias detection. Option C is wrong because Microsoft Defender XDR is a security tool.

Option D is wrong because Azure AI Language detects PII, not bias.

794
MCQhard

You are designing a solution to extract customer names and addresses from scanned handwritten forms. The forms are stored as images in Azure Blob Storage. The extraction must achieve high accuracy with minimal manual review. Which combination of Azure AI services should you use?

A.Azure AI Document Intelligence with prebuilt invoice and receipt models
B.Azure AI Document Intelligence with a custom model trained on handwritten forms
C.Azure AI Language Service with custom Named Entity Recognition (NER)
D.Azure AI Computer Vision with OCR and Azure AI Search
AnswerB

Custom models can be trained on handwriting samples to achieve high accuracy.

Why this answer

Option D is correct because Document Intelligence's custom model can be trained on handwritten forms, and Language Service can be used to post-process entities. However, the best approach is to use Document Intelligence with a custom model trained on handwriting. Option A is wrong because Computer Vision OCR does not handle handwriting well.

Option B is wrong because Language Service alone cannot extract from images. Option C is wrong because Form Recognizer (Document Intelligence) prebuilt models are not optimized for handwriting.

795
Multi-Selectmedium

You are building an agentic solution using Microsoft Semantic Kernel. The agent uses a planner to orchestrate multiple functions. You want to improve the planner's ability to handle complex user requests that involve multiple steps. Which THREE strategies should you implement?

Select 3 answers
A.Limit the number of available functions to reduce planning overhead
B.Enable the planner to ask the user for clarification when the request is ambiguous
C.Create composite functions that encapsulate common multi-step sub-tasks
D.Use a simple, generic prompt to avoid overfitting
E.Provide few-shot examples of multi-step workflows in the planner prompt
AnswersB, C, E

Clarification improves accuracy.

Why this answer

Options B, C, and E are correct. B: Providing few-shot examples helps the planner understand complex tasks. C: Creating higher-level composite functions reduces the number of steps the planner needs to orchestrate.

E: Allowing the planner to ask clarifying questions improves accuracy. Option A is incorrect because a simple prompt reduces capability. Option D is incorrect because limiting functions restricts the planner.

796
Multi-Selectmedium

Which TWO Azure services can be used to perform optical character recognition (OCR) on images?

Select 2 answers
A.Azure Computer Vision Read API
B.Azure Face API
C.Azure Video Indexer
D.Azure Custom Vision
E.Azure Form Recognizer
AnswersA, E

Core OCR service.

Why this answer

A and C are correct. Azure Computer Vision Read API is the primary OCR service. Azure Form Recognizer also uses OCR to extract text from documents.

B is wrong because Custom Vision is for custom image classification. D is wrong because Video Indexer is for video analysis. E is wrong because Face API is for face detection.

797
MCQhard

A company uses Azure AI Language's conversational language understanding (CLU) to build a customer support bot. They want to integrate the bot with Microsoft Teams and need to ensure that user authentication is handled by Microsoft Entra ID. However, users report that the bot sometimes fails to respond when they are not signed into Microsoft Entra ID. What is the most likely cause?

A.The CLU model requires a Microsoft Entra ID token for prediction.
B.The CLU project endpoint is not configured to accept anonymous requests.
C.The bot is not registered in Microsoft Entra ID.
D.The bot's authentication settings in Azure Bot Service require Microsoft Entra ID, but the bot is not passing the token correctly.
AnswerD

The bot may be configured to require Microsoft Entra ID authentication, and if the user is not signed in, the bot cannot respond.

Why this answer

Option C is correct because the bot's authentication settings may require Microsoft Entra ID, but if the bot is configured to expect tokens from Microsoft Entra ID and users are not signed in, the bot cannot authenticate requests. Option A is wrong because the CLU project endpoint does not require authentication for each query; authentication is handled at the bot level. Option B is wrong because the CLU model does not require a Microsoft Entra ID token for prediction.

Option D is wrong because the bot framework itself does not require Microsoft Entra ID; it can use other authentication methods.

798
MCQeasy

You need to monitor the costs of your Azure AI services across multiple subscriptions. Which Azure tool should you use to track spending and set budgets?

A.Azure Cost Management
B.Azure Portal
C.Azure Monitor
D.Azure Advisor
AnswerA

Cost Management provides cost analysis and budgets.

Why this answer

Azure Cost Management is the dedicated tool for monitoring, analyzing, and controlling cloud spending across multiple subscriptions. It provides cost analysis, budget creation, and alerting capabilities specifically designed for tracking Azure AI services costs at scale.

Exam trap

The trap here is that candidates often confuse Azure Monitor (which tracks resource metrics and logs) with cost monitoring, but Azure Monitor has no native capability to track financial spend or set budgets.

How to eliminate wrong answers

Option B is wrong because Azure Portal is the web-based management interface for provisioning and configuring resources, not a dedicated cost tracking and budgeting tool. Option C is wrong because Azure Monitor focuses on performance metrics, logs, and alerts for resource health and application diagnostics, not financial cost tracking. Option D is wrong because Azure Advisor provides best-practice recommendations for optimizing resource usage, security, and reliability, but it does not offer direct cost tracking or budget management features.

799
MCQmedium

A company uses Azure OpenAI to generate marketing copy. They want to ensure that the generated content does not contain offensive language. Which feature should they enable?

A.Use DALL-E to generate images instead of text.
B.Use a system message instructing the model to avoid offensive language.
C.Enable diagnostic logging to review all outputs.
D.Enable content filtering at the deployment level.
AnswerD

Content filtering proactively blocks offensive content.

Why this answer

Option D is correct because Azure OpenAI provides built-in content filtering at the deployment level that automatically detects and blocks offensive or harmful language in both input prompts and generated outputs. This feature uses Microsoft's Responsible AI models to enforce safety policies without requiring custom code or manual review, making it the most reliable and scalable solution for preventing offensive content in marketing copy.

Exam trap

The trap here is that candidates often assume prompt engineering (system messages) is sufficient for safety, but Azure OpenAI requires explicit content filtering at the deployment level to enforce policies reliably and prevent bypassing via prompt injection.

How to eliminate wrong answers

Option A is wrong because DALL-E is an image generation model, not a text filtering mechanism; switching to images does not address the requirement to prevent offensive language in text outputs. Option B is wrong because a system message is a prompt engineering technique that provides guidance to the model but does not guarantee enforcement; the model may still generate offensive content if the instruction is not followed or if the model is manipulated. Option C is wrong because diagnostic logging only records outputs for review after generation, not preventing offensive content in real-time; it is a monitoring tool, not a content filter.

800
Multi-Selecteasy

Which TWO Azure AI services can be used to extract text from images as part of an Azure AI Search enrichment pipeline?

Select 2 answers
A.Azure Bot Service
B.Azure AI Speech to text
C.Azure AI Language translation
D.Azure AI Document Intelligence's read model
E.Azure AI Search's built-in OCR skill
AnswersD, E

Extracts text from documents and images.

Why this answer

B and D are correct. B: The OCR skill in Azure AI Search uses Azure AI Vision. D: Azure AI Document Intelligence uses OCR.

A is for translation; C is for speech; E is for conversational AI.

801
MCQmedium

Refer to the exhibit. You submit this request to the Azure AI Language service. What is the expected response?

A.An error because the request format is invalid.
B.Only key phrase extraction results.
C.Both entity recognition and key phrase extraction results.
D.Only entity recognition results.
AnswerC

The request includes both tasks, so both are executed and results returned.

Why this answer

Option C is correct. The request includes two tasks: entity recognition and key phrase extraction. The response will contain both sets of results.

Option A is wrong because entity recognition is included. Option B is wrong because key phrase extraction is included. Option D is wrong because the request is valid.

802
MCQeasy

A company needs to extract personally identifiable information (PII) from customer support transcripts stored in Azure Blob Storage. Which Azure AI service should they use?

A.Azure AI Speech
B.Azure AI Language Service
C.Azure AI Translator
D.Azure AI Vision
AnswerB

Azure AI Language Service includes PII detection.

Why this answer

The correct answer is B because Azure AI Language Service includes PII detection. A (Azure AI Vision) handles images, not text. C (Azure AI Speech) handles audio.

D (Azure AI Translator) translates text.

803
Multi-Selecteasy

Which TWO Azure AI services can be used to analyze sentiment in text data?

Select 2 answers
A.Azure AI Translator
B.Azure AI Language Service
C.Azure AI Vision
D.Azure AI Content Safety
E.Azure AI Speech
AnswersB, D

It includes sentiment analysis.

Why this answer

The correct answers are A and C. A (Azure AI Language Service) includes sentiment analysis. C (Azure AI Content Safety) can detect harmful content, which includes sentiment aspects.

B (Azure AI Vision) is for images. D (Azure AI Speech) is for audio. E (Azure AI Translator) is for translation.

804
MCQeasy

A developer is creating a custom question answering project in Azure AI Language. The knowledge base contains product manuals in PDF format. Which step is essential before importing the PDFs?

A.Ensure PDFs are in a supported format and accessible
B.Create an Azure AI Search index
C.Deploy a QnA Maker service
D.Translate PDFs to English
AnswerA

PDFs must be in a supported format and accessible via URL or upload.

Why this answer

Option D is correct because PDFs must be in a supported format and accessible via URL or uploaded. Option A is wrong because QnA Maker is deprecated in favor of custom question answering. Option B is wrong because Azure AI Search is not required.

Option C is wrong because translation is not necessary.

805
MCQeasy

You are developing a mobile app that allows users to take a photo of a product and get information about it. The app must identify the product from the image. Which Azure AI service should you use?

A.Azure AI Vision OCR
B.Azure AI Face API
C.Azure AI Custom Vision with image classification
D.Azure AI Custom Vision with object detection
AnswerC

Image classification assigns a label to the entire image, which is suitable for product identification.

Why this answer

The correct answer is Azure AI Custom Vision with image classification. Custom Vision can be trained to recognize products from images. Option A is wrong because object detection is for locating objects, not classifying the entire image.

Option C is wrong because Face API is for faces. Option D is wrong because OCR is for text extraction.

806
MCQhard

You are using Azure AI Studio to deploy a fine-tuned model for code generation. After deployment, you notice that the model returns nonsensical code snippets. You need to diagnose the issue. What should you check first?

A.Verify that the training data is in JSONL format.
B.Test the base model without fine-tuning to compare outputs.
C.Evaluate the model performance on a held-out test dataset.
D.Check the deployment's rate limits and quotas.
AnswerC

Evaluation helps identify if the model learned properly.

Why this answer

Option C is correct because evaluation of the fine-tuned model on a test dataset reveals if it learned correctly. Option A is wrong because training data format issues might cause problems but are not the first diagnostic step after deployment. Option B is wrong because the base model works for many tasks; the fine-tuning likely introduced the issue.

Option D is wrong because rate limits affect throughput, not quality.

807
MCQhard

You deploy a Custom Vision object detection model to classify vehicles. The model works well in good lighting but fails in low-light conditions. What is the most appropriate action?

A.Add images with different lighting conditions to the training set
B.Increase the probability threshold
C.Increase the number of training iterations
D.Use a domain-specific model for vehicles
AnswerA

Including low-light images trains the model to handle such conditions.

Why this answer

Option A is correct because the core issue is a data distribution mismatch: the model was trained primarily on well-lit images and lacks exposure to low-light examples. Adding images with diverse lighting conditions directly addresses this by enriching the training dataset, enabling the model to learn robust features for low-light scenarios. This aligns with the fundamental principle that Custom Vision models are only as good as the training data they receive.

Exam trap

The trap here is that candidates often confuse model performance tuning (threshold, iterations) with data quality issues, mistakenly believing that adjusting hyperparameters can compensate for missing training scenarios.

How to eliminate wrong answers

Option B is wrong because increasing the probability threshold only adjusts the confidence level required to return a prediction; it does not improve the model's ability to detect objects in low light, and may actually reduce recall by filtering out correct but lower-confidence detections. Option C is wrong because increasing the number of training iterations (epochs) on the same dataset does not introduce new visual patterns; it risks overfitting to the existing well-lit images without addressing the low-light deficiency. Option D is wrong because domain-specific models in Custom Vision are pre-trained on generic vehicle images and do not inherently compensate for lighting variations; the problem is not the domain but the lack of representative lighting conditions in the training set.

808
MCQmedium

A company uses Azure Content Moderator to review user-generated images in a social media app. Recently, the team noticed that images containing subtle adult content are not being flagged. What should they do to improve detection without increasing false positives?

A.Increase the moderation thresholds for adult content.
B.Disable the adult classification tier to allow all images to pass through.
C.Configure a human review team using the Review tool to manually inspect flagged content.
D.Retrain the Content Moderator model with additional labeled images of adult content.
AnswerC

Human review can catch subtle content that automated systems miss, and it helps reduce false positives by confirming or overturning automated decisions.

Why this answer

Option C is correct because Azure Content Moderator is designed to work with human review teams via the Review tool to handle edge cases where automated detection fails. By configuring a human review team, flagged images can be manually inspected to catch subtle adult content that the machine learning model misses, without lowering thresholds that would increase false positives. This approach leverages human judgment to improve detection accuracy while maintaining the existing automated moderation settings.

Exam trap

The trap here is that candidates may assume Azure Content Moderator supports custom model retraining (like Custom Vision), but it is a fixed, pre-trained service that cannot be retrained, making human review the only viable option for improving detection without increasing false positives.

How to eliminate wrong answers

Option A is wrong because increasing moderation thresholds would make the model less sensitive, potentially missing even more subtle adult content, not improving detection. Option B is wrong because disabling the adult classification tier would allow all images to pass through without any moderation, completely defeating the purpose of content moderation. Option D is wrong because Azure Content Moderator does not support retraining its pre-built models with custom labeled images; it is a fixed, pre-trained service that cannot be customized with additional training data.

809
MCQhard

You are an Azure AI engineer at Contoso Ltd. The company has an Azure AI solution that uses Azure AI Language to analyze customer feedback. The solution is deployed in the East US region and uses the S0 pricing tier. Recently, the volume of feedback has increased significantly, causing the service to throttle requests. The application logs show HTTP 429 (Too Many Requests) errors during peak hours. The development team has already implemented retry logic with exponential backoff, but the errors persist. You need to recommend a solution to handle the increased load without changing the application code. The solution must minimize cost. What should you do?

A.Upgrade the Azure AI Language resource to a higher pricing tier (e.g., S1) or create additional resources and load balance.
B.Use Azure Front Door to cache responses and reduce load on the service.
C.Switch the Azure AI Language resource to the Free tier to reduce costs.
D.Move the service to a different Azure region with higher capacity.
AnswerA

Higher tiers provide higher rate limits and throughput.

Why this answer

Upgrading to a higher pricing tier (e.g., S1) increases the transactions-per-second (TPS) limit, directly addressing the HTTP 429 throttling errors without requiring code changes. Creating additional resources and load-balancing also distributes the request volume, but upgrading the existing resource is the simplest and most cost-effective approach when retry logic already fails.

Exam trap

The trap here is that candidates may think caching (Azure Front Door) or region relocation can solve throttling, but they fail to recognize that throttling is a rate-limit issue tied to the pricing tier, not network latency or content caching.

How to eliminate wrong answers

Option B is wrong because Azure Front Door caches static content at the edge, but Azure AI Language API responses are dynamic and cannot be cached, so it does not reduce the number of API calls hitting the service. Option C is wrong because the Free tier has extremely low rate limits (e.g., 20 calls per minute) and would immediately throttle even more, making the problem worse. Option D is wrong because moving to a different region does not change the S0 tier's capacity limits; all regions offer the same TPS for a given tier, so throttling would persist.

810
MCQmedium

A developer is building an application to extract text from scanned invoices using Azure Computer Vision's Read API. The invoices contain a mix of printed and handwritten text. The developer needs to ensure the highest accuracy for both types. Which parameter should they set in the API call?

A.Set the 'language' parameter to 'en' for English handwriting.
B.No special parameter; the Read API automatically handles both.
C.Specify the 'model-version' as '2022-04-30'
D.Use the 'mode' parameter set to 'Handwriting'
AnswerB

Read API OCR works on both printed and handwritten text without additional parameters.

Why this answer

The Read API in Azure Computer Vision is designed to extract text from images and documents, and it automatically handles both printed and handwritten text without requiring any special parameter. Setting the 'language' parameter to 'en' is optional and only improves accuracy for language-specific text, but it does not enable or disable handwriting recognition. Therefore, no additional parameter is needed to achieve the highest accuracy for both types.

Exam trap

The trap here is that candidates confuse the Read API with the older OCR API, which had a 'mode' parameter for handwriting, leading them to incorrectly assume a similar parameter is needed in the Read API.

How to eliminate wrong answers

Option A is wrong because the 'language' parameter is used to specify the language of the text for language-specific optimization, but it does not control whether handwriting is recognized; the Read API automatically detects and processes both printed and handwritten text regardless of this parameter. Option C is wrong because specifying a 'model-version' like '2022-04-30' only selects a specific version of the Read API model, but it does not enable or disable handwriting recognition; the latest model versions already support both printed and handwritten text by default. Option D is wrong because the Read API does not have a 'mode' parameter; the 'mode' parameter is a misconception from the older OCR API (Computer Vision OCR), not the Read API, which always processes both printed and handwritten text in a single call.

811
MCQmedium

You are a security engineer for a financial services company. The company uses Azure AI Language to analyze customer communications for compliance. The solution processes sensitive personal data. You need to ensure that all data transmitted to the Azure AI Language service is encrypted in transit and that the service endpoint is not accessible from the public internet. Additionally, you must use Microsoft Entra ID for authentication. The current implementation uses API keys and the public endpoint. You need to reconfigure the solution. What should you do?

A.Configure a private endpoint and continue using the public endpoint for redundancy
B.Disable the public network access without configuring a private endpoint
C.Enable Microsoft Entra ID authentication but keep the public endpoint and API keys
D.Disable the public network access, configure a private endpoint, enable managed identity, and enforce HTTPS
AnswerD

Meets all security requirements.

Why this answer

Option D is correct because it addresses all three requirements: disabling public network access removes internet exposure, configuring a private endpoint ensures traffic stays within the Azure backbone and your virtual network, enabling managed identity allows Microsoft Entra ID authentication without API keys, and enforcing HTTPS guarantees encryption in transit via TLS. This combination fully secures the Azure AI Language service for sensitive personal data.

Exam trap

The trap here is that candidates may think disabling public network access alone is sufficient (Option B), but without a private endpoint, the service becomes unreachable, and they may overlook that managed identity is required to replace API keys for Microsoft Entra ID authentication.

How to eliminate wrong answers

Option A is wrong because continuing to use the public endpoint for redundancy still exposes the service to the public internet, violating the requirement that the endpoint not be accessible from the public internet. Option B is wrong because disabling public network access without a private endpoint leaves no way to connect to the service, as the service would be unreachable. Option C is wrong because keeping the public endpoint and API keys fails to restrict public internet access and does not eliminate the use of API keys, contradicting the requirement to use Microsoft Entra ID authentication exclusively.

812
MCQeasy

A developer is configuring an Azure AI Language resource for sentiment analysis. The solution must process social media posts in real-time with a throughput of 1000 requests per minute. After testing, the developer notices that the API returns a 429 (Too Many Requests) error when the load exceeds 500 requests per minute. What is the most likely cause and solution?

A.Scale out the resource by creating multiple Azure AI Language instances and load balancing requests.
B.Upgrade the Azure AI Language resource to a higher tier (e.g., Standard S) to increase the rate limit.
C.Implement retry logic with exponential backoff to handle 429 errors.
D.Use Azure API Management to cache responses and reduce calls.
AnswerB

The Free tier has a limit of 20 requests per minute; upgrading to Standard S allows up to 1000 requests per minute.

Why this answer

The 429 error indicates the request rate exceeds the resource's allocated tier limit. Azure AI Language resources have predefined rate limits per pricing tier; the Standard S tier offers higher throughput (e.g., 1,000 requests per minute) compared to lower tiers. Upgrading to Standard S directly increases the rate limit to match the required 1,000 requests per minute, making it the correct solution.

Exam trap

The trap here is that candidates often confuse rate limiting with transient errors and choose retry logic (C), not realizing that a consistent 429 at a specific threshold indicates a hard capacity cap that only a tier upgrade can resolve.

How to eliminate wrong answers

Option A is wrong because scaling out with multiple instances and load balancing does not increase the per-instance rate limit; it distributes load but each instance still enforces its own tier-based cap, and the 429 error originates from a single resource's limit being exceeded. Option C is wrong because retry logic with exponential backoff handles transient failures, not capacity limits; it would only delay inevitable errors if the rate consistently exceeds the tier's maximum. Option D is wrong because API Management caching reduces repeated calls for identical responses, but social media posts are unique and uncacheable; caching does not address the fundamental rate limit issue.

813
Multi-Selectmedium

A healthcare organization is implementing a knowledge mining solution to extract information from medical records. They need to ensure that the solution can identify medical conditions, medications, and treatment procedures using a pre-built model. The solution must be deployed in Microsoft Foundry. Which THREE components should be included? (Choose three.)

Select 3 answers
A.Text Analytics for Health skill in an Azure AI Search skillset.
B.Azure AI Search index.
C.Text Analytics for Health model in Microsoft Foundry.
D.Azure AI Document Intelligence (formerly Form Recognizer) custom model.
E.Language Understanding (LUIS) model.
AnswersA, B, C

This skill integrates the model into the indexing pipeline.

Why this answer

Option A is correct because the Text Analytics for Health model is a pre-built model for extracting healthcare entities. Option C is correct because Azure AI Search is used to index the extracted data. Option D is correct because the Text Analytics for Health skill in Azure AI Search applies the model.

Options B and E are incorrect: Language Understanding is for conversational AI, and Form Recognizer is for form data.

814
Drag & Dropmedium

Drag and drop the steps to implement an Azure AI Bot Service with QnA Maker into the correct order.

Drag steps to the numbered slots on the right, or tap a step then tap a slot.

Steps
Order

Why this order

Start with QnA Maker, build the knowledge base, create the bot, connect it, and test.

815
MCQmedium

You are building a chatbot that must understand user intents from free-text input. You have a small set of labeled examples. Which Azure AI Language feature should you use to classify intents with minimal effort?

A.Entity Linking
B.Custom Text Classification
C.Language Detection
D.Conversational Language Understanding (CLU)
AnswerD

CLU is designed for intent classification and entity extraction from conversational utterances.

Why this answer

Conversational Language Understanding (CLU) is the correct service for intent classification from conversational utterances. It supports custom models with minimal data. Custom text classification is for document-level classification, not intents.

Language Detection identifies languages, not intents. Entity Linking extracts entities, not intents.

816
MCQhard

Refer to the exhibit. You are deploying an AI project in Microsoft Foundry using an ARM template. The deployment fails with an error indicating that the hub resource is not in the same region. What is the most likely cause?

A.The resource group for the hub is different from the project's resource group
B.The identity type is SystemAssigned, but the hub requires UserAssigned
C.The hub and project are in different Azure regions
D.The location property is misspelled in the template
AnswerC

Project must be in same region as its hub.

Why this answer

In Azure AI Foundry, the hub and project resources must reside in the same Azure region because the project is a child resource of the hub and relies on the hub's regional endpoint for metadata storage and compute orchestration. The ARM template deployment fails with a region mismatch error when the location property of the project specifies a different region than the hub's location, as Azure enforces regional affinity for child resources to ensure low-latency communication and data residency compliance.

Exam trap

The trap here is that candidates confuse resource group boundaries with regional boundaries, assuming that resources in different resource groups cannot be linked, when in fact Azure allows cross-resource-group parent-child relationships as long as the regions match.

How to eliminate wrong answers

Option A is wrong because the resource group can be different for the hub and project; Azure allows child resources to be in a separate resource group from the parent, so a resource group mismatch does not cause a region-related deployment failure. Option B is wrong because the identity type (SystemAssigned vs. UserAssigned) is unrelated to regional constraints; the hub does not require a specific identity type for region validation, and identity configuration affects authentication, not deployment location.

Option D is wrong because a misspelled location property would typically result in a validation error (e.g., 'Invalid template property') rather than a specific 'not in the same region' error; the error message explicitly indicates a regional mismatch, not a syntax issue.

817
Multi-Selectmedium

Which TWO Azure AI services can be used to implement a content moderation pipeline that detects hate speech and blocks violent images?

Select 2 answers
A.Azure AI Content Safety
B.Azure AI Translator
C.Azure AI Speech to Text
D.Azure AI Language sentiment analysis
E.Azure AI Vision Image Analysis
AnswersA, E

Detects hate speech and other offensive text.

Why this answer

Azure AI Content Safety is correct because it is specifically designed to detect hate speech, profanity, and other harmful text content, while Azure AI Vision Image Analysis can analyze images for violent or inappropriate visual content. Together, they form a comprehensive content moderation pipeline that addresses both text and image moderation requirements.

Exam trap

The trap here is that candidates may confuse Azure AI Language sentiment analysis (which only measures positive/negative sentiment) with Azure AI Content Safety (which specifically detects hate speech and harmful content), or assume Azure AI Translator can handle moderation tasks because it processes text.

818
MCQmedium

You are using Azure OpenAI Service to generate code snippets for a development team. You notice that the generated code sometimes contains security vulnerabilities. You need to minimize the risk of generating insecure code while maintaining productivity. What should you do?

A.Use system messages to instruct the model to prioritize security
B.Fine-tune the model on a dataset of secure code
C.Set the temperature parameter to 0
D.Disable content filtering to allow more flexibility
AnswerA

System messages set behavior and can guide the model to generate secure code.

Why this answer

Option D is correct because system messages set the behavior of the model, instructing it to prioritize security. Option A is wrong because disabling content filtering would not address security vulnerabilities. Option B is wrong because fine-tuning requires labeled data and may not be practical.

Option C is wrong because temperature affects creativity, not security.

819
MCQmedium

You are an AI developer at a legal firm. The firm uses Azure AI Language to extract entities from legal documents. The current custom NER model is trained on a small dataset and has low accuracy for certain entity types like 'Statute' and 'Case Citation'. You need to improve the model's performance without increasing the labeling effort significantly. You have the following options: Option A: Add more labeled examples for the underperforming entity types by manually labeling additional documents. Option B: Use the prebuilt entity recognition model from Azure AI Language and map its outputs to custom entities. Option C: Enable active learning in the custom NER project and review the suggested labels from the model. Option D: Train a new model using the Azure Machine Learning automated ML (AutoML) for text classification.

A.Option D
B.Option B
C.Option A
D.Option C
AnswerD

Active learning focuses effort on uncertain predictions.

Why this answer

Option C is correct. Active learning suggests uncertain predictions for manual review, focusing effort on the most impactful examples. Option A is wrong because it increases labeling effort significantly.

Option B is wrong because prebuilt models may not extract legal-specific entities. Option D is wrong because AutoML is for classification, not NER entity extraction.

820
MCQmedium

You are building a solution to analyze images of handwritten medical prescriptions. The text is in English and includes drug names and dosages. Which combination of Azure AI services should you use?

A.Azure AI Computer Vision Read API and Azure AI Language
B.Azure AI Custom Vision and Azure AI Language
C.Azure AI Document Intelligence and Azure AI Language
D.Azure AI Video Indexer and Azure AI Language
AnswerA

Read API extracts text, Language Service extracts entities.

Why this answer

Option C is correct because the Computer Vision Read API extracts handwritten text, and the Language Service can then extract entities like drug names. Document Intelligence (A) is for forms, not free-text prescriptions. Custom Vision (B) is not suitable.

Video Indexer (D) is for video.

821
MCQmedium

You are designing an Azure AI solution for a global e-commerce company. The solution must: (1) Translate product descriptions into 12 languages in real-time. (2) Detect sentiment in customer reviews for each language. (3) Extract key product attributes (e.g., color, size) from unstructured review text. (4) Store results in a centralized database for analytics. The solution must minimize latency and cost. You plan to use Azure AI services. Which combination of services should you use?

A.Azure AI Language for translation and sentiment, and Azure AI Computer Vision for attribute extraction.
B.Azure AI Speech for translation, Azure AI Language for sentiment, and Azure AI Search for storage.
C.Azure AI Translator for translation, Azure AI QnA Maker for attribute extraction, and Azure SQL Database for storage.
D.Azure AI Translator for translation, Azure AI Language for sentiment and entity extraction, and Azure AI Search to index the results.
AnswerD

Translator provides real-time translation, Language provides sentiment and entity extraction, and Search stores results for analytics.

Why this answer

Option D is correct because Azure AI Translator provides real-time translation into 12 languages, Azure AI Language handles both sentiment analysis and entity extraction (for attributes like color and size) from unstructured text, and Azure AI Search indexes the results for low-latency analytics. This combination minimizes latency by using dedicated services for each task and avoids unnecessary overhead from services like Speech or Computer Vision that are not required for text-only processing.

Exam trap

The trap here is that candidates may confuse Azure AI Language's entity extraction with Azure AI Computer Vision or QnA Maker, assuming attribute extraction requires visual analysis or Q&A logic, when in fact it is a text-based NLP task handled by the Language service.

How to eliminate wrong answers

Option A is wrong because Azure AI Computer Vision is designed for image analysis, not for extracting attributes from unstructured text; attribute extraction from text is a job for Azure AI Language's entity extraction. Option B is wrong because Azure AI Speech is for speech-to-text and text-to-speech, not for text translation, and Azure AI Search is a search indexer, not a storage solution; storage should be a database like Cosmos DB or SQL Database. Option C is wrong because Azure AI QnA Maker is for building question-answer bots, not for extracting product attributes from review text; attribute extraction requires entity recognition, which Azure AI Language provides.

822
MCQhard

Your Azure OpenAI application experiences high latency during peak hours. You have already scaled up the deployment to the maximum PTUs. What is the most effective next step to reduce latency?

A.Create multiple deployments across different regions and use Azure Traffic Manager to distribute requests
B.Use Azure OpenAI's global deployment with the same PTU
C.Switch from GPT-4 to GPT-3.5-turbo
D.Increase the token limit per request
AnswerA

Geographic load balancing spreads load and reduces latency.

Why this answer

When PTU deployment is already maxed out, the bottleneck is the capacity of a single regional deployment. Distributing requests across multiple regional deployments via Azure Traffic Manager (using performance or geographic routing) spreads the load, reducing per-deployment contention and lowering latency. This approach leverages regional redundancy and global load balancing without requiring a model change or sacrificing quality.

Exam trap

The trap here is that candidates assume 'global deployment' (Option B) provides automatic load distribution, but in reality it still uses a single PTU pool and does not distribute load across regions; the correct approach is to explicitly create multiple regional deployments and route traffic with a traffic manager.

How to eliminate wrong answers

Option B is wrong because a global deployment with the same PTU still routes all traffic through a single quota pool and regional endpoint, so it does not alleviate the capacity bottleneck during peak hours. Option C is wrong because switching to a less capable model reduces quality and may not address the root cause of high latency if the deployment is already saturated; it is a workaround, not a scaling solution. Option D is wrong because increasing the token limit per request actually increases processing time per request, worsening latency under load, and does not increase throughput capacity.

823
Multi-Selectmedium

Which TWO Azure AI Language features can you use to extract structured data from unstructured text?

Select 2 answers
A.Sentiment Analysis
B.Language Detection
C.Entity Linking
D.Key Phrase Extraction
E.Named Entity Recognition
AnswersC, E

Entity Linking provides structured links to known entities.

Why this answer

Named Entity Recognition (NER) extracts entities like people, places. Entity Linking disambiguates entities by linking to a knowledge base. Key Phrase Extraction extracts key topics but not structured data.

Sentiment Analysis gives sentiment. Language Detection identifies language.

824
MCQhard

You are using Azure AI Foundry to fine-tune a GPT-3.5 model on a dataset of customer service conversations. The fine-tuning job fails with an error indicating that the training data format is invalid. What is the most likely issue?

A.The training data is not in JSONL format with the correct structure.
B.The training data is in CSV format instead of JSON.
C.The training data contains only one conversation example.
D.The training data does not include the assistant's responses.
AnswerA

Fine-tuning requires JSONL format with each line containing a valid 'messages' array.

Why this answer

Azure AI Foundry requires fine-tuning data to be in JSONL format with a specific structure: each line must be a JSON object containing a 'messages' array with 'role' and 'content' fields for system, user, and assistant turns. The error indicates the training data format is invalid, and the most likely cause is that the data is not in this required JSONL structure, as JSONL is the only accepted format for GPT-3.5 fine-tuning in Azure OpenAI Service.

Exam trap

The trap here is that candidates confuse the general requirement for 'JSON format' with the specific requirement for 'JSONL format with a messages array,' leading them to incorrectly select CSV or plain JSON as the issue, when the real problem is the lack of the correct conversational structure.

How to eliminate wrong answers

Option B is wrong because CSV format is not supported for fine-tuning GPT-3.5 models in Azure AI Foundry; the service requires JSONL, not JSON or CSV, and CSV lacks the nested 'messages' structure needed for conversational data. Option C is wrong because having only one conversation example does not cause a format error; it may lead to poor model performance but the format itself would still be valid if structured correctly. Option D is wrong because while missing assistant responses would make the data unusable for training, the error specifically indicates a format issue, not a content issue; the JSONL structure could still be technically valid without assistant responses.

825
MCQhard

You are building a generative AI solution using Azure OpenAI Service. The application must retrieve information from a large private knowledge base. You need to ensure the model uses only relevant documents from the knowledge base to generate answers. Which feature should you configure?

A.Implement a custom prompt flow
B.Use Azure OpenAI On Your Data with vector search
C.Configure a content filter
D.Fine-tune the model with the knowledge base
AnswerB

This feature enables retrieval-augmented generation (RAG) using vector search.

Why this answer

B is correct because Azure OpenAI On Your Data with vector search enables the model to retrieve only the most semantically relevant documents from a private knowledge base by converting both the user query and the documents into high-dimensional vectors and performing similarity search. This ensures the model's responses are grounded in the specific, relevant information without exposing the entire knowledge base to the model.

Exam trap

The trap here is that candidates often confuse fine-tuning (D) with retrieval-augmented generation (RAG), assuming that training the model on the knowledge base is the best way to ground answers, when in fact RAG with vector search is the correct pattern for dynamic, relevant document retrieval without modifying the base model.

How to eliminate wrong answers

Option A is wrong because implementing a custom prompt flow does not inherently include a retrieval mechanism; it only orchestrates the sequence of calls and prompts, so it cannot ensure that only relevant documents are used from the knowledge base. Option C is wrong because configuring a content filter is a safety mechanism to block harmful or inappropriate content, not a retrieval or grounding feature to select relevant documents. Option D is wrong because fine-tuning the model with the knowledge base would bake the entire knowledge into the model's weights, which is inefficient, costly, and does not allow dynamic retrieval of only relevant documents per query; it also risks overfitting and cannot handle updates to the knowledge base without retraining.

Page 10

Page 11 of 14

Page 12