Microsoft Azure AI Engineer Associate AI-102 (AI-102) — Questions 526600

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

Page 7

Page 8 of 14

Page 9
526
MCQmedium

You have configured an Azure AI Search indexer with a Cosmos DB data source as shown in the exhibit. The indexer runs successfully, but you notice that the index is missing some documents that were recently added to Cosmos DB. What is the most likely cause?

A.The indexer is not configured to track changes using _ts.
B.The container name is misspelled.
C.The high water mark is not being updated correctly, causing some documents to be skipped.
D.The query does not select all fields required by the index.
AnswerC

If the high water mark is not updated, documents with _ts <= high water mark are skipped.

Why this answer

The query uses '@HighWaterMark' which is a placeholder for the high water mark value used for change tracking. However, the query includes 'WHERE c._ts > @HighWaterMark' which filters out documents with a timestamp less than or equal to the high water mark. If the high water mark is not being updated correctly or if documents have the same timestamp, they might be missed.

Option A is wrong because the change tracking is enabled by the query using _ts. Option B is wrong because the query selects all fields needed. Option C is wrong because the container name is correct.

Option D is wrong because the connection string is valid since the indexer runs successfully.

527
Multi-Selecteasy

You are using Azure OpenAI Service to generate code snippets. The output must be safe and free of security vulnerabilities. Which TWO practices should you follow? (Select TWO.)

Select 2 answers
A.Increase the temperature parameter to encourage diversity
B.Rely on the model's built-in safety features
C.Use Azure AI Content Safety to filter outputs
D.Fine-tune the model with a dataset of secure code examples
E.Include a system message instructing the model to follow secure coding practices
AnswersC, E

Content safety can detect and block harmful code.

Why this answer

Option C is correct because Azure AI Content Safety is a dedicated service that provides an additional layer of filtering for harmful or inappropriate content, including security vulnerabilities, beyond what the model itself offers. It allows you to define custom severity thresholds and blocklists, ensuring that generated code snippets are safe before they reach the user. This is a recommended practice for production deployments to mitigate risks like injection attacks or exposure of sensitive patterns.

Exam trap

The trap here is that candidates often assume the model's built-in safety features are sufficient (Option B) or that increasing temperature (Option A) improves safety by adding randomness, when in fact Azure AI Content Safety is the explicit, exam-tested tool for output filtering in generative AI solutions.

528
Multi-Selecthard

Which THREE components are required to build a custom named entity recognition (NER) model in Azure Cognitive Service for Language?

Select 3 answers
A.A QnA Maker knowledge base for entity definitions.
B.A LUIS application to handle entity extraction.
C.A set of labeled documents for training.
D.A language service project with a custom NER schema.
E.A deployed endpoint for the custom NER model.
AnswersC, D, E

Labeled data is essential for training.

Why this answer

Option C is correct because a custom NER model in Azure Cognitive Service for Language requires a set of labeled documents for training. These labeled documents define the entities and their spans within text, which the model uses to learn patterns for extraction. Without labeled data, the model cannot be trained to recognize custom entities.

Exam trap

The trap here is that candidates often confuse the separate Azure AI services (QnA Maker, LUIS, Language service) and assume they are interchangeable for custom NER, when in fact each has a distinct role and pipeline.

529
Multi-Selecthard

Which THREE components are required to build a custom chat application using Azure OpenAI Service that can answer questions based on your own private data?

Select 3 answers
A.The 'Add your data' feature configured in Azure OpenAI Studio.
B.Azure AI Search index.
C.An Azure OpenAI Service deployment.
D.A fine-tuned custom model.
E.An Azure OpenAI embeddings model deployment.
AnswersA, B, C

Enables grounding on private data.

Why this answer

Option A is correct because the 'Add your data' feature in Azure OpenAI Studio provides a no-code interface to connect your private data sources (e.g., Azure Blob Storage, local files) to an Azure OpenAI chat model. It automatically chunks the data, creates an Azure AI Search index, and configures the retrieval-augmented generation (RAG) pipeline, enabling the model to answer questions grounded in your proprietary content without fine-tuning.

Exam trap

The trap here is that candidates often confuse fine-tuning (Option D) with retrieval-augmented generation, assuming that custom data requires model retraining, when in fact the 'Add your data' feature uses a RAG approach that does not modify the base model.

530
Matchingmedium

Match each Azure AI term to its definition.

Drag a concept onto its matching description — or click a concept then click the description.

Concepts
Matches

Language Understanding Intelligent Service

Service to create a question and answer bot

Service to build custom image classifiers

Convert spoken language to text

Extract insights from text like key phrases

Why these pairings

These are key terms and services in Azure AI.

531
MCQmedium

A healthcare organization uses Azure AI Vision to analyze medical images for abnormalities. The solution must ensure patient data is not stored in the service's logs. What should you configure?

A.Use a private endpoint to access the Azure AI Vision resource.
B.Set the logging level to 'None' in the Azure AI Vision resource diagnostic settings.
C.Enable Customer-Managed Keys (CMK) for the resource.
D.Use Azure Policy to restrict data residency.
AnswerB

Disabling logging prevents data from being stored.

Why this answer

Option A is correct because Azure AI services allow you to disable logging of request and response data to meet data privacy requirements. Option B is wrong because network isolation does not prevent logging. Option C is wrong because data residency controls where data is stored, not logging.

Option D is wrong because encryption protects data at rest, not logging.

532
MCQmedium

Your company wants to build a custom generative AI model that generates architectural designs. The model should be trained on the company's proprietary dataset of floor plans and designs. Which Azure service should you use?

A.Azure OpenAI Service
B.Azure Machine Learning
C.Azure AI Vision
D.Azure AI Document Intelligence
AnswerB

Azure Machine Learning supports training custom generative models using your own data.

Why this answer

Option A is correct because Azure Machine Learning provides a platform for training custom models, including generative models. Option B is wrong because Azure OpenAI Service offers pre-trained models, not custom training. Option C is wrong because Azure AI Document Intelligence is for extracting information from documents.

Option D is wrong because Azure AI Vision is for image analysis, not generative design.

533
MCQhard

You have deployed a generative AI model using Azure Machine Learning. The model is used for generating financial reports. You need to monitor the model's performance and detect data drift in the input data. What should you use?

A.Azure Machine Learning data drift monitoring
B.Azure Monitor
C.Application Insights
D.Azure AI Language
AnswerA

This feature monitors changes in input data distribution compared to training data, alerting to drift.

Why this answer

Option D is correct because Azure Machine Learning data drift monitoring is designed to detect changes in input data over time. Option A is wrong because Azure Monitor is for infrastructure monitoring. Option B is wrong because Application Insights is for application telemetry.

Option C is wrong because Azure AI Language is for NLP tasks.

534
MCQmedium

You are troubleshooting an agentic solution where the agent is not returning responses within acceptable time limits. You suspect the agent is making too many sequential calls to external tools. Which strategy should you recommend to reduce latency?

A.Increase the max token limit
B.Enable parallel tool execution
C.Add more tools to distribute the load
D.Reduce the thread history length
AnswerB

Parallel execution allows the agent to invoke multiple tools simultaneously.

Why this answer

Parallel tool execution reduces latency by running independent tool calls concurrently. Adding more tools increases load. Increasing token limit does not help with tool call latency.

Reducing thread history might affect context but not tool call speed.

535
MCQeasy

A healthcare organization uses Azure AI Language to extract medical entities from clinical notes. The solution must comply with HIPAA. What should you configure?

A.Use Azure AI Language with HIPAA-enabled service and enable encryption
B.Configure data residency to a specific region
C.Enable content filtering
D.Isolate the service in a virtual network
AnswerA

Azure AI Language offers HIPAA-compliant tiers with data encryption.

Why this answer

Option A is correct because Azure AI Language can be configured with a HIPAA-eligible service tier, and enabling encryption at rest and in transit ensures compliance with HIPAA's security and privacy rules. This configuration meets the healthcare organization's requirement to protect protected health information (PHI) when extracting medical entities from clinical notes.

Exam trap

The trap here is that candidates often confuse network isolation (VNet) or data residency with HIPAA compliance, not realizing that HIPAA mandates specific encryption and a signed BAA, which are only available with a HIPAA-eligible service tier.

How to eliminate wrong answers

Option B is wrong because configuring data residency to a specific region addresses data sovereignty but does not by itself ensure HIPAA compliance, which requires additional safeguards like encryption and access controls. Option C is wrong because content filtering in Azure AI Language is designed to block offensive or harmful content, not to enforce HIPAA compliance or protect PHI. Option D is wrong because isolating the service in a virtual network (VNet) provides network-level security but does not automatically meet HIPAA requirements; encryption and a HIPAA-eligible service tier are still necessary.

536
MCQeasy

Your company uses Azure AI Content Safety to moderate user-generated content in a chat application. You need to detect and block sexual content in multiple languages. Which pre-built category should you configure?

A.Sexual
B.Self-harm
C.Hate
D.Violence
AnswerA

This category specifically detects sexual content.

Why this answer

Azure AI Content Safety provides pre-built severity-based categories for content moderation. The 'Sexual' category is specifically designed to detect and block explicit sexual content, including text and images, across multiple languages. This makes it the correct choice for your requirement to moderate sexual content in a chat application.

Exam trap

The trap here is that candidates may confuse 'Sexual' with broader categories like 'Hate' or 'Violence', not realizing that Azure AI Content Safety has a dedicated pre-built category for sexual content with specific detection capabilities.

How to eliminate wrong answers

Option B (Self-harm) is wrong because it focuses on content related to self-injury or suicide, not sexual material. Option C (Hate) is wrong because it targets hate speech based on protected attributes like race or religion, not sexual content. Option D (Violence) is wrong because it detects violent acts or threats, which are distinct from sexual content.

537
MCQeasy

You need to generate realistic synthetic data using Azure OpenAI Service to train a machine learning model. The data must be diverse and cover edge cases. Which approach should you use?

A.Use prompt engineering with detailed instructions to generate varied examples.
B.Fine-tune the model on a small dataset of real examples.
C.Use Azure OpenAI embeddings to generate similar data points.
D.Set a high temperature parameter only.
AnswerA

Prompt engineering effectively controls output diversity and coverage.

Why this answer

Option A is correct because prompt engineering with detailed descriptions can guide the model to generate diverse data covering edge cases. Option B is wrong because fine-tuning on existing data may reduce diversity. Option C is wrong because temperature alone is not sufficient; prompt engineering is key.

Option D is wrong because embeddings don't generate data.

538
MCQhard

You are using Azure AI Search to index a set of PDF documents. The index includes a 'content' field with the extracted text. Users report that when they search for 'budget forecast', documents containing only 'budget' or 'forecast' are ranked lower than expected. Which configuration change would improve the ranking for multi-word queries?

A.Add a separate field for each word in the document
B.Change the analyzer to a custom analyzer that splits on spaces only
C.Enable semantic search on the index
D.Set the 'content' field to a higher boosting value
AnswerC

Semantic search uses advanced ranking models that consider the meaning and relationship between words.

Why this answer

Option B is correct because enabling semantic search improves ranking by understanding the context of multi-word queries. Option A is wrong because adding more fields does not directly improve ranking for multi-word queries. Option C is wrong because setting a higher boost on 'content' does not fix the issue of missing proximity.

Option D is wrong because changing the analyzer to a different language may not help.

539
Multi-Selecthard

You are deploying a solution that uses Azure OpenAI Service to generate financial reports. You need to ensure the outputs are accurate and consistent. Which TWO parameters should you adjust? (Choose two.)

Select 2 answers
A.Set presence_penalty to 0.5.
B.Set temperature to 0.
C.Set max_tokens to 2000.
D.Set frequency_penalty to 0.7.
E.Set top_p to 0.1.
AnswersB, E

Low temperature makes output more deterministic.

Why this answer

A and C are correct. Temperature set to 0 reduces randomness for consistent outputs. Top_p set to 0.1 forces high-probability tokens.

B is wrong because presence_penalty encourages topic diversity, not consistency. D is wrong because frequency_penalty reduces repetition, which might be desired but not for consistency. E is wrong because max_tokens controls length, not accuracy or consistency.

540
MCQeasy

You are using Azure AI Document Intelligence to extract data from purchase orders. The purchase orders have a table of line items. Which prebuilt model should you use?

A.Prebuilt invoice model
B.Prebuilt document model
C.Layout model
D.Custom model
AnswerA

Invoice model extracts tables and item lines.

Why this answer

Option B is correct because the prebuilt invoice model extracts tables and line items. Option A is wrong because the general document model extracts text and layout but not invoice-specific fields. Option C is wrong because the custom model requires training data.

Option D is wrong because the layout model only extracts text and tables without field mapping.

541
MCQhard

A financial services firm uses Azure AI Language to analyze earnings call transcripts. They need to extract key phrases and identify sentiment for each speaker's turn. Which approach should they use?

A.Call the prebuilt sentiment analysis API on the entire transcript
B.Split the transcript by speaker turns and call key phrase extraction and sentiment analysis on each part
C.Use QnA Maker to extract Q&A pairs per speaker
D.Use Text Analytics for health to extract entities and sentiment
AnswerB

Splitting by speaker turns allows per-speaker analysis.

Why this answer

Option C is correct because the CLU entity linking can link mentions to entities, but for speaker-level sentiment, they need to process each utterance separately. Option A is wrong because QnA Maker is for FAQ. Option B is wrong because prebuilt sentiment analysis does not differentiate speakers.

Option D is wrong because Text Analytics for health is for healthcare, not financial.

542
MCQhard

Refer to the exhibit. You are reviewing a content safety policy for an Azure AI Foundry deployment. The policy rate limits to 20 requests per minute. A user submits 50 requests in one minute. How many requests are allowed?

A.20
B.50
C.100
D.None, all are blocked.
AnswerA

The rate limit policy sets 20 requests per minute.

Why this answer

The content safety policy enforces a rate limit of 20 requests per minute. When a user submits 50 requests in one minute, the rate limiter allows only the first 20 requests and blocks the remaining 30. This is a standard token-bucket or sliding-window rate-limiting behavior in Azure AI Foundry, where exceeding the limit results in HTTP 429 (Too Many Requests) for excess requests.

Exam trap

Microsoft often tests the misconception that exceeding a rate limit blocks all requests, when in fact the limit is a threshold that allows the first N requests and denies the rest.

How to eliminate wrong answers

Option B is wrong because it assumes all 50 requests are allowed, ignoring the explicit rate limit of 20 per minute. Option C is wrong because 100 is not related to any limit in the policy; it may confuse the rate limit with a burst or quota value. Option D is wrong because the policy does not block all requests; it allows up to the limit (20) and then blocks the excess.

543
MCQeasy

You are using Azure AI Search to index customer support tickets. You want to automatically extract the customer's sentiment and key phrases from each ticket. Which Azure AI service should you integrate as a skillset?

A.Azure AI Document Intelligence
B.Azure AI Computer Vision
C.Azure AI Translator
D.Azure AI Language
AnswerD

Offers sentiment and key phrase extraction skills.

Why this answer

Option A is correct because Azure AI Language provides sentiment analysis and key phrase extraction as built-in skills. Option B is wrong because Document Intelligence is for document extraction. Option C is wrong because Translator is for translation.

Option D is wrong because Computer Vision is for images.

544
MCQeasy

A company is building a chatbot that must handle user queries in multiple languages. The chatbot uses Azure AI Language Service. Which feature should be used to detect the language of incoming messages before routing them to the appropriate language model?

A.Sentiment Analysis
B.Key Phrase Extraction
C.Language Detection
D.Entity Recognition
AnswerC

Language Detection identifies the language of text input.

Why this answer

The correct answer is A because Language Detection is the feature specifically designed to identify the language of text. B (Key Phrase Extraction) extracts key phrases without language detection; C (Sentiment Analysis) determines sentiment; D (Entity Recognition) identifies entities like names and locations.

545
MCQeasy

A company uses Azure Face API to verify employee identities for building access. They need to ensure that only live faces are used, not photos or videos. Which feature should they enable?

A.Set a high confidence threshold for face matching.
B.Face identification with a large person group.
C.Enable liveness detection using session-based verification.
D.Face detection with attributes such as age and emotion.
AnswerC

Liveness detection checks for spoofing attacks.

Why this answer

Option C is correct because Azure Face API's liveness detection with session-based verification is specifically designed to prevent spoofing attacks using photos, videos, or masks. It analyzes subtle cues such as micro-movements, texture, and depth to confirm the presence of a live person, ensuring that only live faces are accepted for identity verification.

Exam trap

The trap here is that candidates may confuse confidence thresholds or face attributes with liveness detection, not realizing that only session-based verification actively checks for spoofing through motion and depth analysis.

How to eliminate wrong answers

Option A is wrong because setting a high confidence threshold only increases the strictness of face matching scores, but does not differentiate between a live face and a spoofed image or video. Option B is wrong because face identification with a large person group is used to match a detected face against a database of enrolled persons, but it does not verify liveness or detect presentation attacks. Option D is wrong because face detection with attributes like age and emotion extracts demographic and emotional information from a face, but it cannot determine whether the face is live or a reproduction.

546
MCQmedium

An organization uses Microsoft Copilot Studio to create an agent for IT support. The agent should be able to reset passwords, unlock accounts, and look up user information by connecting to on-premises Active Directory via Microsoft Entra ID. Which type of authentication should be used for the agent to access these actions?

A.Basic authentication with username and password.
B.API key authentication to the on-premises API.
C.Certificate-based authentication for the agent's identity.
D.OAuth 2.0 authentication with Microsoft Entra ID using on-behalf-of flow.
AnswerD

On-behalf-of flow allows the agent to obtain tokens to call downstream APIs on behalf of the user.

Why this answer

Option A is correct because OAuth 2.0 with on-behalf-of flow allows the agent to authenticate to Microsoft Entra ID and then access on-premises resources via an API. Option B is wrong because API key is not suitable for user-specific actions. Option C is wrong because certificate-based authentication is not typical for this scenario.

Option D is wrong because basic authentication is insecure and deprecated.

547
MCQhard

You are a generative AI engineer at a financial services company. The company uses Azure OpenAI Service to generate investment summaries. You have deployed a GPT-4 model with a content filter set to 'Low' for hate speech. The model frequently generates summaries that include biased language against certain demographics. You need to reduce biased outputs while maintaining the ability to generate detailed financial analysis. You cannot afford to retrain the model. You have the following options: A) Change the content filter severity to 'High' for all categories, B) Add a system message instructing the model to avoid bias and provide examples of unbiased summaries in the prompt, C) Use the Azure AI Language service to detect bias in the output and regenerate if bias is found, D) Deploy a different model like GPT-3.5 which has less bias. Which course of action should you take?

A.Use the Azure AI Language service to detect bias in the output and regenerate if bias is found.
B.Add a system message instructing the model to avoid bias and provide examples of unbiased summaries in the prompt.
C.Change the content filter severity to 'High' for all categories.
D.Deploy a different model like GPT-3.5 which has less bias.
AnswerB

This approach guides the model to produce unbiased outputs by providing explicit instructions and examples.

Why this answer

Option B is correct because adding a system message and examples in the prompt (few-shot learning) can effectively reduce biased outputs without retraining. Option A is wrong because increasing content filter severity may block legitimate financial analysis content. Option C is wrong because detecting bias after generation and regenerating is inefficient and may still produce biased output.

Option D is wrong because GPT-3.5 may also exhibit bias and may not provide the same quality of financial analysis.

548
MCQeasy

You are developing a custom chatbot using Azure AI Bot Service and Language Understanding (CLU). The chatbot needs to escalate to a human agent when the user's sentiment is negative. Which component should you use to detect sentiment?

A.Azure AI Language sentiment analysis
B.Azure Cognitive Search
C.Orchestration workflow
D.QnA Maker
AnswerA

Sentiment analysis detects positive/negative sentiment.

Why this answer

Option B is correct because the Azure AI Language sentiment analysis is a prebuilt feature that can be integrated into the bot. Option A is wrong because Orchestration workflow routes intents, not sentiment. Option C is wrong because QnA Maker is for FAQs.

Option D is wrong because Azure Cognitive Search is for indexing.

549
MCQhard

You are implementing a knowledge mining solution for a legal firm. The solution must ingest large volumes of legal documents (PDFs and Word files) stored in Azure Blob Storage. You need to extract text, recognize named entities (e.g., parties, judges, case numbers), and index the content for full-text search. The solution should also support redaction of sensitive information before indexing. Which combination of Azure AI services should you use?

A.Azure AI Document Intelligence, Azure AI Translator, and Azure AI Search
B.Azure AI Document Intelligence, Azure AI Video Indexer, and Azure AI Search
C.Azure AI Document Intelligence, Azure AI Language, custom skill for redaction, and Azure AI Search
D.Azure AI Document Intelligence, Azure AI Content Safety, and Azure AI Search
AnswerC

Document Intelligence extracts text, Language recognizes entities, custom skill redacts, Search indexes.

Why this answer

Azure AI Document Intelligence extracts text from documents. Azure AI Language provides entity recognition. A custom skill can perform redaction.

Azure AI Search indexes the content. Option A is wrong because Azure AI Translator is not needed. Option B is wrong because Azure AI Video Indexer is for video.

Option D is wrong because Azure AI Content Safety is for moderation, not redaction.

550
MCQeasy

You are a data engineer at a university. The university wants to digitize its historical student records (paper forms) to make them searchable. The records are scanned as images (JPEG) and stored in Azure Blob Storage. Each form contains handwritten fields: student name, ID number, date of birth, and degree. You need to extract these fields and index them in Azure AI Search. The solution must use Azure AI Services and minimize manual labeling effort. Which approach should you take?

A.Use Azure AI Custom Vision to train a model to detect handwriting regions, then use Azure AI Vision OCR to read text.
B.Use Azure AI Search with a blob indexer and a skillset that includes OCR skill and Entity Recognition skill.
C.Use Azure AI Document Intelligence to train a custom extraction model with a few labeled samples, then deploy as a custom skill in Azure AI Search.
D.Use Azure AI Vision OCR to extract text from images, then use Azure AI Language to extract entities like name, date, and degree.
AnswerC

Document Intelligence is designed for extraction from forms with minimal labeling.

Why this answer

Option B is correct because Azure AI Document Intelligence has prebuilt models for handwriting and can extract fields with minimal training. Option A requires custom training and labeling. Option C uses OCR but not extraction.

Option D uses Custom Vision which is not suitable for text extraction.

551
Multi-Selecteasy

You are developing an Azure AI solution that uses pre-built models from Azure AI Vision to analyze images. The solution must be able to detect objects and read printed text. Which TWO capabilities should you use?

Select 2 answers
A.OCR (legacy)
B.Facial detection
C.Object detection
D.Image tagging
E.Read (OCR)
AnswersC, E

Detects objects in images.

Why this answer

Option C is correct because Azure AI Vision's Object Detection capability identifies and locates objects within an image, returning bounding box coordinates and labels. This directly meets the requirement to 'detect objects' in the solution.

Exam trap

The trap here is that candidates often confuse Image Tagging (which only provides labels) with Object Detection (which provides both labels and spatial localization), and may mistakenly choose the legacy OCR API instead of the modern Read API for text extraction.

552
MCQmedium

You are working for a healthcare organization that uses Azure AI Document Intelligence to process patient intake forms. The forms are scanned and uploaded as multi-page PDFs. The extraction accuracy for the 'diagnosis code' field is poor. You have a labeled dataset of 200 forms. You need to improve the extraction accuracy without writing custom code. The solution must also handle forms with varying layouts. What should you do?

A.Use the 'Form processing' custom extraction model.
B.Use the US Tax W-2 predefined model as a base and customize it.
C.Use the General Document model to extract all text and then parse.
D.Create a custom neural model and train it with the labeled dataset.
AnswerD

Custom neural models handle varied layouts and improve field accuracy.

Why this answer

Option C is correct because custom neural models are designed for varied layouts and can be trained with labeled data. Option A is wrong because predefined models are for fixed layouts. Option B is wrong because the form processing feature is for key-value pairs, not free-form fields.

Option D is wrong because the general document model is less accurate for specific fields.

553
Multi-Selecthard

Which THREE considerations are important when planning to use Azure OpenAI Service in a production environment?

Select 3 answers
A.Configure content filtering to block harmful outputs
B.Train a custom model from scratch using your own data
C.Plan for rate limits and quotas to handle expected load
D.Determine data residency requirements for your region
E.Deploy the service within a virtual network (VNet)
AnswersA, C, D

Required for responsible use.

Why this answer

Option A is correct because Azure OpenAI Service includes built-in content filtering to detect and block harmful outputs such as hate, violence, or self-harm. This is a critical safety and compliance requirement for production deployments, as it helps meet responsible AI principles and regulatory obligations. Without configuring content filters, the service could generate inappropriate responses that violate usage policies or legal standards.

Exam trap

Microsoft often tests the misconception that Azure OpenAI Service allows training models from scratch, but the service only supports fine-tuning of pre-trained base models, not full custom training.

554
MCQeasy

You are implementing a generative AI solution using Azure OpenAI. You need to ensure that the model's outputs do not contain certain inappropriate words or phrases. Which feature should you configure?

A.System message instructions
B.Grounding with your data
C.Content filters
D.Max tokens limit
AnswerC

Content filters can block inappropriate words and phrases.

Why this answer

Content filters in Azure OpenAI are specifically designed to detect and block inappropriate words or phrases in both prompts and completions. They operate at the service level, applying configurable severity thresholds for categories like hate, violence, sexual content, and self-harm, ensuring model outputs adhere to policy without requiring prompt engineering or data modifications.

Exam trap

Microsoft often tests the misconception that system messages (Option A) are sufficient for content safety, when in fact they are only behavioral guidelines and lack the enforcement mechanism of dedicated content filters.

How to eliminate wrong answers

Option A is wrong because system message instructions guide model behavior and tone but cannot reliably enforce content restrictions; they are advisory and can be overridden by the model, especially in edge cases. Option B is wrong because grounding with your data (using Azure Cognitive Search) augments prompts with your own data for relevance and accuracy, but it does not filter or block inappropriate content from the model's generated responses. Option D is wrong because the max tokens limit controls the length of the output, not its content; it cannot prevent the model from generating inappropriate words or phrases within the allowed token count.

555
Multi-Selecthard

You are designing an agentic solution using Azure AI Agent Service. The agent needs to perform actions on behalf of users, such as sending emails and updating databases. The solution must use managed identities for authentication to Azure resources. Which TWO configurations are required?

Select 2 answers
A.Store connection strings in Azure Key Vault and reference them in the agent's configuration
B.Create a service principal in Microsoft Entra ID and assign RBAC roles to the agent's resource
C.Use DefaultAzureCredential in the agent's code to authenticate to Azure services
D.Configure the agent to use an API key for each external service
E.Assign a system-assigned managed identity to the Azure resource hosting the agent
AnswersC, E

DefaultAzureCredential uses managed identity.

Why this answer

Options B and D are correct. B: The agent must have a managed identity assigned to authenticate to resources like Microsoft Graph and Azure SQL Database. D: The agent's code must use the managed identity to acquire tokens, typically via DefaultAzureCredential.

Option A is incorrect because connection strings expose secrets. Option C is incorrect because RBAC permissions are assigned to the managed identity, not the agent's service principal directly. Option E is incorrect because connection strings are not used with managed identities.

556
MCQmedium

A developer uses the Azure OpenAI API to generate code. They want to ensure that the generated code is in Python. Which parameter should they set?

A.temperature
B.top_p
C.system message
D.max_tokens
AnswerC

System message can guide the model to output Python code.

Why this answer

The system message is used to set the behavior and context of the AI model, including specifying the desired output format or language. By setting the system message to 'You are a helpful assistant that always writes code in Python', the developer can instruct the model to generate Python code consistently. This parameter is part of the chat completions API and directly influences the model's persona and constraints.

Exam trap

Microsoft often tests the distinction between parameters that control output randomness (temperature, top_p) and those that control output structure or behavior (system message), leading candidates to mistakenly choose temperature or top_p for language specification.

How to eliminate wrong answers

Option A is wrong because temperature controls the randomness of the output, not the language or format of the generated code. Option B is wrong because top_p (nucleus sampling) controls the cumulative probability threshold for token selection, which affects diversity but does not specify the output language. Option D is wrong because max_tokens limits the length of the generated response, not the programming language or content type.

557
MCQmedium

A company wants to generate personalized product descriptions for its e-commerce site using Azure OpenAI. They need to ensure the model's output adheres to brand guidelines and does not generate prohibited content. Which approach should they use?

A.Use a system message with brand guidelines and apply content filtering.
B.Use prompt engineering with negative prompts and ignore content filtering.
C.Provide few-shot examples in the user message and rely on the model's training.
D.Fine-tune the model with brand guidelines and disable content filtering for performance.
AnswerA

System messages set behavior, content filtering blocks prohibited content.

Why this answer

Option A is correct because using a system message allows you to embed brand guidelines directly into the conversation context, instructing the model on tone, style, and prohibited content. Azure OpenAI's content filtering provides an additional safety layer by automatically detecting and blocking harmful or policy-violating outputs, ensuring compliance with both brand and regulatory requirements.

Exam trap

Microsoft often tests the misconception that fine-tuning or prompt engineering alone is sufficient for safety and compliance, when in reality Azure OpenAI requires explicit content filtering and system messages to enforce brand guidelines reliably.

How to eliminate wrong answers

Option B is wrong because ignoring content filtering removes the safety guardrails that prevent prohibited content, and negative prompts alone are unreliable for enforcing brand guidelines. Option C is wrong because few-shot examples in the user message do not guarantee consistent adherence to brand guidelines across all outputs, and relying solely on the model's training ignores the need for explicit content filtering. Option D is wrong because disabling content filtering for performance sacrifices safety and compliance, and fine-tuning alone cannot dynamically enforce brand guidelines as effectively as a system message combined with content filtering.

558
MCQeasy

You want to use the Azure AI Language service to summarize long customer support conversations into a short summary. Which feature should you use?

A.Sentiment Analysis
B.Conversational Summarization
C.Entity Extraction
D.Key Phrase Extraction
AnswerB

Generates a summary of conversations with multiple participants.

Why this answer

Option C is correct because Conversational Summarization is designed to summarize conversations between multiple participants. Option A is wrong because Key Phrase Extraction extracts key phrases, not a summary. Option B is wrong because Entity Extraction identifies entities.

Option D is wrong because Sentiment Analysis determines sentiment, not a summary.

559
MCQmedium

Refer to the exhibit. { "content_filters": [ { "type": "hate", "action": "block", "severity": "high" }, { "type": "sexual", "action": "block", "severity": "medium" }, { "type": "self_harm", "action": "block", "severity": "low" } ] } You deploy an Azure OpenAI model with the above content filter configuration. A user submits a prompt that the system rates as "hate" at severity level "medium". What happens?

A.The prompt is allowed because the severity is below the threshold.
B.The prompt is blocked because hate content is detected.
C.The prompt is blocked because the severity is medium.
D.The prompt is allowed because the hate filter is not configured for medium.
AnswerA

The hate filter blocks only at high severity.

Why this answer

Option C is correct because the filter for hate blocks only at severity 'high', so a 'medium' severity hate message is allowed. Options A and B are wrong because the action is block only for high severity. Option D is wrong because the filter is not skipped entirely.

560
MCQeasy

You are a developer for a healthcare startup. They are building a mobile app that allows users to take photos of prescription labels and extract medication names, dosages, and frequencies. The app must run on iOS and Android devices. The solution should use a pre-built AI service with minimal custom code. What should you recommend?

A.Use Azure AI Document Intelligence (Form Recognizer) to analyze labels
B.Train a Custom Vision object detection model to locate and read text
C.Use Azure AI Language to extract entities from the label text
D.Use Azure AI Vision OCR API to extract text from the label images
AnswerD

Pre-built OCR works for text extraction from natural scenes.

Why this answer

Azure AI Vision OCR provides pre-built text extraction from images. Custom Vision requires custom training. Form Recognizer is for forms.

Azure AI Language is for text analytics, not image text extraction.

561
Drag & Dropmedium

Drag and drop the steps to build and deploy a custom Azure AI Document Intelligence model 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

First gather and label documents, create the resource, train the model, test, and publish.

562
MCQhard

Your company uses Azure OpenAI Service to generate product descriptions. You need to ensure that the generated content does not include offensive language and adheres to responsible AI principles. What should you implement?

A.Enable customer-managed key encryption
B.Configure content filters in Azure OpenAI
C.Fine-tune the model with a curated dataset
D.Set usage limits and throttling
AnswerB

Content filters block offensive language and support responsible AI.

Why this answer

Content filters in Azure OpenAI allow you to define categories (e.g., hate, violence, self-harm) and severity levels (low, medium, high) to automatically block or flag offensive language in generated outputs. This directly enforces responsible AI principles by preventing harmful content from being surfaced to users, without requiring model retraining or encryption changes.

Exam trap

The trap here is that candidates often confuse data security controls (like encryption or throttling) with content safety controls, assuming any 'security' feature can filter offensive language, when in fact only purpose-built content filters can analyze and block harmful text in real time.

How to eliminate wrong answers

Option A is wrong because customer-managed key encryption (CMK) protects data at rest but does not inspect or filter the semantic content of model outputs for offensive language. Option C is wrong because fine-tuning with a curated dataset can reduce but not guarantee the absence of offensive outputs; it cannot dynamically block real-time content violations and requires ongoing dataset maintenance. Option D is wrong because usage limits and throttling control API request rates and quotas, not the quality or safety of the generated text.

563
MCQmedium

You are planning a multi-region deployment of Azure AI services to ensure high availability and low latency for global users. You need to decide how to manage API keys and endpoint URLs across regions. What is the recommended approach?

A.Create separate Azure AI services resources per region and hardcode the keys in application configuration.
B.Create a single Azure AI services resource in one region and share the key across all applications.
C.Use Azure API Management to expose a single endpoint that routes to regional Azure AI services resources, each with its own key stored in Key Vault.
D.Use Microsoft Entra ID authentication with managed identities and have each application call the region-specific endpoint directly.
AnswerC

API Management provides routing, failover, and key management.

Why this answer

Option C is correct because it combines Azure API Management as a unified gateway with regional Azure AI Services resources, each secured by individual keys stored in Azure Key Vault. This architecture provides global load balancing, regional failover, and centralized key management without exposing keys in application code. API Management can route requests based on latency or geography, ensuring high availability and low latency while Key Vault rotates keys securely.

Exam trap

The trap here is that candidates often assume a single global resource or direct regional calls are sufficient, overlooking the need for a centralized gateway like API Management to handle routing, failover, and key management at scale.

How to eliminate wrong answers

Option A is wrong because hardcoding keys in application configuration violates security best practices (keys can be exposed in source control or logs) and requires manual updates per region for key rotation or failover, undermining operational efficiency. Option B is wrong because a single resource in one region creates a single point of failure and introduces cross-region latency for users far from that region, violating the high-availability and low-latency requirements. Option D is wrong because while Microsoft Entra ID with managed identities is secure, calling region-specific endpoints directly requires the application to manage regional routing logic and failover itself, adding complexity and defeating the purpose of a unified gateway for global load balancing.

564
MCQmedium

Your application needs to extract key phrases from customer reviews to identify common topics. Which Azure AI Language feature should you use?

A.Sentiment Analysis
B.Language Detection
C.Named Entity Recognition
D.Key Phrase Extraction
AnswerD

Key Phrase Extraction identifies main talking points.

Why this answer

Key Phrase Extraction is the correct feature for extracting key points from text. Entity Recognition extracts named entities. Sentiment Analysis determines sentiment.

Language Detection identifies language.

565
MCQmedium

Your organization uses Microsoft 365 Copilot. You want to ensure that Copilot only uses data from your Microsoft 365 tenant and does not access external sources. Which setting should you configure?

A.Disable 'Allow Copilot to use external data' in the Copilot settings.
B.Configure Copilot to use only Microsoft Graph data.
C.Enable content filtering for Copilot.
D.Disable the Bing search integration in the Microsoft 365 admin center.
AnswerB, D

Copilot by default uses Graph; no extra config needed.

Why this answer

Option B is correct because configuring Copilot to use only Microsoft Graph data ensures that the AI model retrieves content exclusively from your Microsoft 365 tenant (e.g., emails, documents, calendar events) via Microsoft Graph APIs. This setting explicitly restricts Copilot from querying external sources like the public web or third-party services, aligning with the requirement to keep data within the tenant boundary.

Exam trap

The trap here is that candidates confuse disabling Bing search integration (Option D) with fully restricting Copilot to tenant-only data, but Bing integration only controls web search, not other external data sources like public Microsoft Graph endpoints or third-party connectors.

How to eliminate wrong answers

Option A is wrong because there is no setting named 'Allow Copilot to use external data' in Microsoft 365 Copilot; the actual control is through the 'Microsoft Search' or 'Bing search integration' settings. Option C is wrong because content filtering controls the moderation of harmful or sensitive content in Copilot responses, not the scope of data sources Copilot can access. Option D is wrong because disabling Bing search integration only prevents Copilot from using Bing as a search engine for web results, but it does not restrict Copilot from accessing other external sources like third-party connectors or public data via Microsoft Graph; the correct approach is to limit data sources to Microsoft Graph only.

566
MCQeasy

You are designing an Azure AI solution that uses Azure AI Language to analyze customer support transcripts. The solution must identify key phrases, detect sentiment, and extract custom entities specific to your product catalog. Which two Azure AI Language features should you enable?

A.PII Detection
B.Key Phrase Extraction
C.Summarization
D.Custom Entity Extraction
AnswerB, D

Extracts key topics and terms from text.

Why this answer

Key Phrase Extraction (Option B) is correct because it identifies the main points and important terms in customer support transcripts, such as 'refund request' or 'account issue,' which directly supports analyzing the content. Custom Entity Extraction (Option D) is correct because it allows you to define and extract domain-specific entities from your product catalog, such as product names or model numbers, using a trained custom entity extraction model. Together, these two features enable both general insight extraction and tailored, product-specific data extraction from the transcripts.

Exam trap

Microsoft often tests the distinction between pre-built features (like Key Phrase Extraction) and custom features (like Custom Entity Extraction), and the trap here is that candidates may incorrectly choose Summarization or PII Detection because they sound relevant to 'analyzing transcripts,' but they do not fulfill the specific requirements of key phrase identification and custom entity extraction.

How to eliminate wrong answers

Option A is wrong because PII Detection is designed to identify and redact personally identifiable information (e.g., names, phone numbers, credit card numbers) for privacy compliance, not to analyze key phrases or extract custom product entities. Option C is wrong because Summarization generates a concise summary of the transcript's main points, which is useful for overview but does not perform key phrase identification or custom entity extraction as required by the question.

567
MCQeasy

You run the above Azure CLI command. What is the expected output?

A.The primary and secondary keys along with the endpoint
B.The primary and secondary keys
C.A list of endpoints for the service
D.An error because the command is incorrect
AnswerA, B

The command returns keys; the endpoint is not included in the output of this command.

Why this answer

The Azure CLI command `az cognitiveservices account keys list` retrieves the API keys for a Cognitive Services account. By default, this command returns both the primary and secondary keys, but does not include the endpoint URL. The endpoint must be retrieved separately using `az cognitiveservices account show`.

Therefore, the expected output is only the primary and secondary keys, not the endpoint.

Exam trap

The trap here is that candidates often assume the `keys list` command also returns the endpoint, confusing it with the `show` command that returns the full resource properties including the endpoint.

How to eliminate wrong answers

Option A is wrong because the command does not return the endpoint; it only returns the keys. Option C is wrong because the command returns keys, not a list of endpoints. Option D is wrong because the command is syntactically correct and will execute successfully.

568
MCQmedium

Your company uses Azure AI Document Intelligence to process invoices. You need to extract the invoice date and total amount. Which model should you use?

A.Read model
B.Prebuilt invoice model
C.Layout model
D.General document model
AnswerB

The invoice model is optimized for invoice fields.

Why this answer

The prebuilt invoice model is specifically trained to extract fields like invoice date and total amount from invoices. The layout model extracts text and tables, not specific fields; the general document model is for key-value pairs; the read model only extracts text.

569
MCQeasy

You are deploying a real-time translation service using Azure AI Translator. The solution must support automatic language detection and translation for customer chat conversations. Which pricing tier should you select to minimize costs while meeting the requirement?

A.Free (F0)
B.Standard S3
C.Standard S2
D.Standard S1
AnswerD

S1 provides real-time translation with language detection at moderate cost.

Why this answer

The Standard S1 tier (D) is correct because it provides up to 1 million characters per month for translation and language detection, which is sufficient for typical customer chat workloads. The Free tier (F0) is limited to 2,000 characters per request and 2 million characters per month, which may be insufficient for production chat volumes. Higher tiers (S2, S3) offer increased throughput and character limits but incur higher costs, making S1 the most cost-effective choice that meets the requirement.

Exam trap

The trap here is that candidates often assume the Free tier is sufficient for production workloads, overlooking its strict character and throughput limits, or they choose a higher tier like S2 or S3 thinking more capacity is always better, without considering cost optimization for the actual workload.

How to eliminate wrong answers

Option A is wrong because the Free tier (F0) has a maximum throughput of 2,000 characters per request and a monthly cap of 2 million characters, which is too restrictive for real-time customer chat conversations that may involve high volume or burst traffic. Option B is wrong because Standard S3 offers up to 1 billion characters per month and higher throughput, which is overprovisioned and unnecessarily expensive for a typical chat translation service. Option C is wrong because Standard S2 provides up to 10 million characters per month, which exceeds the needs of most chat workloads and results in higher costs than S1 without additional benefit for this scenario.

570
Multi-Selectmedium

Which TWO Azure services can be used to monitor and analyze the usage costs of an Azure AI solution that includes multiple Cognitive Services accounts?

Select 2 answers
A.Azure Service Health
B.Azure Policy
C.Azure Cost Management + Billing
D.Azure Advisor
E.Azure Monitor metrics and logs
AnswersC, E

Provides cost analysis, budgets, and alerts.

Why this answer

Azure Cost Management + Billing (C) provides native tools to track, analyze, and optimize cloud spending across all Azure services, including multiple Cognitive Services accounts. It allows you to break down costs by resource, resource group, or tag, and set budgets with alerts. Azure Monitor metrics and logs (E) can capture custom metrics and diagnostic logs from Cognitive Services, enabling you to correlate usage patterns with cost data for deeper analysis.

Exam trap

The trap here is that candidates often confuse Azure Advisor's cost recommendations with actual cost monitoring, or mistakenly think Azure Service Health covers billing issues, when in fact only Cost Management + Billing and Azure Monitor provide the direct data needed for usage cost analysis.

571
MCQhard

Your company uses Azure AI Language to analyze customer feedback. The solution currently uses the default endpoint and key. Security policy requires that all API calls be authenticated using Microsoft Entra ID and that network access be restricted to a specific virtual network. You need to reconfigure the resource. What should you do?

A.Disable local authentication, enable managed identity, and configure the virtual network
B.Regenerate the API keys and update the application
C.Create a private endpoint for the resource
D.Configure the resource to use Microsoft Entra ID authentication only
AnswerA

Managed identity uses Microsoft Entra ID, and VNet integration restricts network access.

Why this answer

Option A is correct because it addresses both security requirements: disabling local authentication ensures that API keys cannot be used, forcing all calls to authenticate via Microsoft Entra ID; enabling a managed identity provides a secure identity for the resource to authenticate with Entra ID; and configuring the virtual network restricts network access to the specified VNet, meeting the network restriction policy.

Exam trap

The trap here is that candidates often confuse 'private endpoint' (which only handles network isolation) with the combined requirement of authentication and network access, leading them to select Option C without realizing that local authentication must also be disabled.

How to eliminate wrong answers

Option B is wrong because regenerating API keys and updating the application still relies on local authentication (API keys), which does not satisfy the requirement to use Microsoft Entra ID for authentication. Option C is wrong because creating a private endpoint only restricts network access to a virtual network (via a private IP), but it does not disable local authentication or enforce Microsoft Entra ID authentication; API keys would still be accepted. Option D is wrong because configuring the resource to use Microsoft Entra ID authentication only (via the 'Authentication type' setting) disables local authentication but does not restrict network access to a specific virtual network; network-level controls must be configured separately.

572
MCQhard

You deploy an agent in Microsoft Foundry that uses a custom skill in Azure AI Search. The skill calls an Azure Function to enrich documents. The function uses an API key. The deployment succeeds but the skill returns an error when processing documents. The function logs show the request is received but the API key is missing. What is the most likely cause?

A.The Azure Function is not running.
B.The skill definition does not include the apiKey in the header.
C.The API key stored in Azure Key Vault is expired.
D.The custom skill's URI is incorrect.
AnswerB

The skill must pass the key as a header.

Why this answer

Option B is correct because the skill might not be passing the API key as a header. Option A is wrong because the function is reachable. Option C is wrong because the key might be correct but not passed.

Option D is wrong because the function URL might be correct.

573
MCQeasy

You are a content moderator for a social media platform that uses Azure Content Moderator. The platform has a custom blocklist of URLs (e.g., 'example.com/spam') and a custom term list for hate speech. Recently, users have been posting comments that contain a new form of hate speech not yet in the term list. The comments are being allowed through moderation. You need to update the solution to catch these new phrases as quickly as possible. What should you do?

A.Create a new custom model using Custom Vision to detect the new phrases in text
B.Retrain the image classification model using the new phrases as training data
C.Add the new phrases to the existing custom term list using the List Management API
D.Delete the existing term list and recreate it with the new phrases included
AnswerC

This quickly adds new terms to be matched against incoming comments.

Why this answer

Option C is correct because Azure Content Moderator's custom term lists allow you to dynamically add new offensive terms or phrases via the List Management API, which immediately updates the moderation screening without retraining or redeploying any model. This provides the fastest way to catch new hate speech patterns as they emerge, as the term list is checked in real-time during content review.

Exam trap

The trap here is that candidates may assume retraining or creating a new model is required for new patterns, but Azure Content Moderator's term lists are designed for rapid, rule-based updates without the overhead of model training, and the List Management API enables immediate addition of terms to an existing list.

How to eliminate wrong answers

Option A is wrong because Custom Vision is designed for image classification, not text phrase detection, and cannot be used to identify hate speech in text comments. Option B is wrong because image classification models are irrelevant to text-based hate speech; retraining such a model would not affect text moderation. Option D is wrong because deleting and recreating the term list is unnecessary and slower; the List Management API supports adding new terms to an existing list without disruption, preserving any existing terms and avoiding downtime.

574
MCQhard

You are implementing a knowledge mining solution using Azure AI Search. The data source is a large Azure Cosmos DB collection containing customer support tickets. Each ticket has fields: ticket_id, description, category, and resolution. You need to ensure that the search index can support fuzzy search and autocomplete suggestions. What should you configure in the index definition?

A.Set the 'searchable' attribute on the description field and define a suggester
B.Set the 'filterable' attribute on the description field
C.Set the 'sortable' attribute on the ticket_id field
D.Set the 'facetable' attribute on the category field
AnswerA

Searchable enables full-text search; suggester enables autocomplete.

Why this answer

Option B is correct because fuzzy search requires 'searchable' and 'analyzer' fields; autocomplete requires 'suggestions' configured on a field. Options A, C, D are not correct for these features.

575
MCQhard

You are designing a solution that uses Azure AI Document Intelligence to extract data from invoices. The invoices are in various formats (PDF, TIFF, and JPEG) and languages. You need to ensure high accuracy for extraction. Which approach should you take?

A.Use the prebuilt invoice model and configure language detection.
B.Use Azure AI Vision OCR to extract all text and then parse using regular expressions.
C.Convert all invoices to a single format (e.g., PDF) before processing.
D.Use a custom extraction model trained on a sample set of invoices covering all formats and languages.
AnswerD

Custom models adapt to specific variations.

Why this answer

Option D is correct because Azure AI Document Intelligence’s custom extraction models are trained on your specific invoice samples, enabling the model to learn the unique layouts, fields, and languages present across PDF, TIFF, and JPEG formats. This approach directly addresses the need for high accuracy on varied formats and languages, as the model adapts to the exact patterns in your data rather than relying on a generic prebuilt model or brittle post-processing.

Exam trap

The trap here is that candidates assume the prebuilt invoice model (Option A) is always the best choice for invoices, but the question explicitly requires high accuracy across varied formats and languages, which demands a custom model tailored to the specific data.

How to eliminate wrong answers

Option A is wrong because the prebuilt invoice model supports a fixed set of fields and languages; while language detection can be configured, the model may not achieve high accuracy for invoices with non-standard layouts, rare languages, or format-specific artifacts (e.g., TIFF compression noise). Option B is wrong because Azure AI Vision OCR extracts raw text without understanding invoice structure; parsing with regular expressions is fragile and cannot reliably handle varied layouts, missing fields, or multilingual content, leading to low accuracy. Option C is wrong because converting all invoices to a single format (e.g., PDF) does not improve extraction accuracy—the underlying content and layout remain the same, and conversion can introduce artifacts (e.g., loss of resolution in TIFF-to-PDF) that degrade OCR quality.

576
MCQhard

You are deploying a custom Named Entity Recognition (NER) model using Azure AI Language. You have 500 labeled documents. After training, the model shows high precision but low recall. Which action is most likely to improve recall?

A.Switch to Conversational Language Understanding
B.Reduce the confidence threshold for entity extraction
C.Add more labeled examples covering the missed entities
D.Increase the number of training epochs
AnswerC

More diverse examples help the model generalize and catch more true entities.

Why this answer

Low recall means many true entities are missed. Adding more labeled examples that include the missed entities helps the model learn to recognize them. Increasing training epochs may lead to overfitting.

Changing the model to a different type (e.g., CLU) is not appropriate for NER. Reducing the confidence threshold would allow more predictions but could lower precision, and the question targets recall specifically.

577
Multi-Selecthard

An agent uses Azure AI Language to perform sentiment analysis on customer feedback. The team notices that the sentiment scores are sometimes inaccurate for negative feedback. Which TWO improvements should the team consider?

Select 2 answers
A.Pre-process the text to handle negations and sarcasm.
B.Use a custom sentiment analysis model trained on domain-specific data.
C.Switch to a different language model without fine-tuning.
D.Increase the number of decimal places in the sentiment score.
E.Increase the confidence threshold for positive sentiment.
AnswersA, B

Pre-processing can improve sentiment detection.

Why this answer

Option A is correct because Azure AI Language's pre-built sentiment analysis models can struggle with linguistic nuances like negations (e.g., 'not good') and sarcasm (e.g., 'Great, another delay'). Pre-processing the text to explicitly handle these patterns—such as by expanding contractions or using negation detection—can improve the accuracy of the sentiment scores before they are passed to the model.

Exam trap

The trap here is that candidates often assume that increasing precision or switching models generically will fix inaccuracies, rather than recognizing that domain-specific fine-tuning and text pre-processing are the standard Azure AI Language approaches to handle linguistic edge cases like negations and sarcasm.

578
MCQmedium

A company is building a chatbot using Azure AI Language. The chatbot must understand user intents and extract entities like dates and locations. The solution should minimize manual labeling effort. Which feature should the team use?

A.Conversational Language Understanding (CLU) Orchestration Workflow
B.CLU prebuilt intents
C.CLU prebuilt entity components
D.Azure AI QnA Maker
AnswerC

Prebuilt entity components recognize common entities like dates and locations without manual labeling.

Why this answer

Option C is correct because the CLU prebuilt entity components can automatically recognize common entities like dates and locations without custom labeling. Option A is wrong because Orchestration Workflow routes to different projects but does not extract entities. Option B is wrong because prebuilt intents cover common intents, not entities.

Option D is wrong because QnA Maker is for FAQ, not intent/entity extraction.

579
MCQeasy

You run the PowerShell script shown to audit your Azure AI Agent Service agents. The script outputs that several agents have no tools configured. What is the impact on those agents?

A.The agents cannot be deployed until tools are added
B.The agents can only respond to queries using the model's built-in knowledge, without ability to perform actions
C.The agents cannot start conversations with users
D.The agents will use default tools provided by Azure
AnswerB

No tools means no external actions.

Why this answer

Option A is correct because without tools, the agent can only respond based on its instructions and model, limiting its capabilities. Option B is incorrect because the agent can still start conversations. Option C is incorrect because the agent can still be deployed.

Option D is incorrect because the agent can still be invoked.

580
Multi-Selecteasy

Which TWO actions are valid ways to authenticate to Azure AI services?

Select 2 answers
A.Use a client certificate.
B.Use an API key.
C.Use Microsoft Entra ID authentication with a service principal.
D.Use a shared access signature (SAS) token.
E.Use a managed identity.
AnswersB, C

API keys are a common authentication method.

Why this answer

Option B is correct because Azure AI services accept API keys as a straightforward authentication method. Each service generates a pair of keys that must be included in the `Ocp-Apim-Subscription-Key` header of HTTP requests. Option C is correct because Microsoft Entra ID (formerly Azure AD) authentication with a service principal is fully supported for Azure AI services, allowing token-based authentication via the `Authorization` header with a bearer token obtained from the Microsoft identity platform.

Exam trap

The trap here is that candidates confuse managed identities as a direct authentication method for Azure AI services, when in fact they are an identity provisioning mechanism that requires an additional token exchange step to authenticate via Entra ID.

581
Multi-Selectmedium

Which TWO Azure services can be used to implement a conversational AI solution that understands user intent and responds appropriately?

Select 2 answers
A.Azure Bot Service
B.Conversational Language Understanding
C.Azure AI Speech-to-Text
D.Azure AI Translator
E.Azure AI Search
AnswersA, B

Bot Service allows building conversational bots.

Why this answer

Azure Bot Service provides a framework for building bots. Conversational Language Understanding (CLU) extracts intents. Azure AI Search is for indexing/search.

Translator is for translation. Speech-to-Text is for transcription.

582
Multi-Selecthard

Which THREE components are required to build a custom copilot using Microsoft Copilot Studio that can answer questions from a SharePoint document library?

Select 3 answers
A.A Microsoft Copilot Studio copilot
B.A knowledge source configured in Copilot Studio (e.g., Azure Cognitive Search)
C.A Power Automate flow to trigger the copilot
D.A SharePoint site with the documents
E.An Azure OpenAI Service deployment
AnswersA, B, D

The copilot is the conversational interface.

Why this answer

Options A, C, and E are correct. A copilot is needed as the interface. SharePoint must be configured as a data source.

A knowledge source (cognitive search) enables indexing and retrieval. Option B is wrong because Azure OpenAI is not required; Copilot Studio uses its own AI. Option D is wrong because Power Automate is not required for basic Q&A.

583
MCQmedium

You are creating an agent in Microsoft Copilot Studio that needs to escalate to a human agent when it cannot resolve a query. Which feature should you use?

A.Add a 'Transfer to agent' topic.
B.Add an 'Escalate' system topic.
C.Configure the 'Fallback' topic to call a Power Automate flow.
D.Use the 'End conversation' node.
AnswerA

Transfers to human agent.

Why this answer

In Microsoft Copilot Studio, the 'Transfer to agent' topic is the correct feature to escalate unresolved queries to a human agent. This topic is specifically designed to hand off the conversation to a live agent, often by triggering a handoff mechanism such as a Dynamics 365 Customer Service queue or a custom integration. It ensures that the bot gracefully transfers context and conversation history, maintaining a seamless user experience.

Exam trap

The trap here is that candidates confuse the 'Escalate' system topic (which does not exist) with the 'Transfer to agent' topic, or they mistakenly think the 'Fallback' topic can handle escalation, when in fact it is only for unrecognized input and not for intentional handoffs.

How to eliminate wrong answers

Option B is wrong because the 'Escalate' system topic does not exist in Copilot Studio; the correct system topic for escalation is the 'Transfer to agent' topic, which is a built-in topic that can be customized. Option C is wrong because the 'Fallback' topic is used to handle unrecognized user input, not to escalate to a human agent; calling a Power Automate flow from it could trigger external actions but does not inherently provide a human handoff mechanism. Option D is wrong because the 'End conversation' node simply terminates the bot session without any escalation, leaving the user without assistance.

584
MCQhard

A company uses Azure OpenAI to generate product descriptions. They notice that the model occasionally produces descriptions that include false claims about product features. The company needs to reduce the frequency of these inaccuracies without changing the training data. Which parameter adjustment would be most effective?

A.Increase the top_p parameter
B.Increase the max_tokens parameter
C.Decrease the temperature parameter
D.Increase the frequency_penalty parameter
AnswerC

Lower temperature makes the model more focused and less likely to hallucinate.

Why this answer

Decreasing the temperature parameter reduces the randomness of the model's output, making it more deterministic and less likely to generate creative but factually incorrect statements. This directly addresses the need to reduce false claims without modifying training data, as lower temperature forces the model to rely on its most probable (and typically more accurate) token predictions.

Exam trap

The trap here is that candidates often confuse temperature with creativity or length control, assuming that increasing randomness (higher temperature) or extending output length (max_tokens) will somehow improve accuracy, when in fact lower temperature is the standard parameter for reducing hallucinations.

How to eliminate wrong answers

Option A is wrong because increasing top_p (nucleus sampling) expands the set of candidate tokens considered, which increases output diversity and can actually worsen factual inaccuracies by allowing less probable tokens. Option B is wrong because increasing max_tokens only extends the maximum length of the generated text; it does not influence the factual accuracy or creativity of the content. Option D is wrong because increasing frequency_penalty penalizes tokens that have already appeared, reducing repetition but not addressing the root cause of hallucinated or false claims.

585
MCQmedium

You are developing a chatbot using Azure AI Bot Service that uses Azure OpenAI Service for natural language understanding. The chatbot must be deployed in multiple regions for low latency. However, the customer requires that all customer data remain within the European Union. Which architecture should you recommend?

A.Deploy the bot in a single EU region and rely on CDN for static content.
B.Use Azure Front Door to route traffic to a single bot hosted in West Europe.
C.Deploy the bot and all supporting services in multiple EU regions, e.g., West Europe and North Europe, without cross-region data replication.
D.Deploy the bot in multiple global regions and use Cosmos DB with multi-region writes.
AnswerC

Data stays within EU; multiple regions provide low latency.

Why this answer

Option C is correct because deploying the bot and all supporting services in multiple EU regions (e.g., West Europe and North Europe) ensures low latency for users across Europe while keeping all customer data within the European Union. Azure OpenAI Service and Azure AI Bot Service can be deployed regionally without cross-region data replication, satisfying the data residency requirement. This architecture avoids any data leaving EU boundaries while providing geographic redundancy for performance.

Exam trap

The trap here is that candidates often assume multi-region deployment automatically requires cross-region data replication (like Cosmos DB multi-region writes), but the correct approach is to deploy independent regional instances without replicating data across regions to satisfy strict data residency requirements.

How to eliminate wrong answers

Option A is wrong because relying on a CDN for static content does not address the core requirement of low latency for dynamic chatbot interactions, and a single-region deployment still creates a single point of failure for the bot logic and AI processing. Option B is wrong because using Azure Front Door to route traffic to a single bot hosted in West Europe does not provide true multi-region deployment; the bot itself remains in one region, so users far from West Europe will experience higher latency, and data still resides only in that one EU region, failing to optimize latency across multiple EU locations. Option D is wrong because deploying the bot in multiple global regions and using Cosmos DB with multi-region writes would replicate customer data outside the European Union, violating the data residency requirement; additionally, multi-region writes in Cosmos DB replicate data across regions, which is not permitted under the EU-only constraint.

586
MCQeasy

You are deploying a generative AI solution using Azure OpenAI Service. You need to monitor the usage and costs associated with the service. What should you use?

A.Microsoft Purview
B.Azure Advisor
C.Azure Cost Management
D.Azure Monitor
AnswerD

Azure Monitor provides metrics and logs for monitoring usage and costs.

Why this answer

Azure Monitor is the correct choice because it provides detailed metrics, logs, and alerts for Azure OpenAI Service usage, including token consumption, request counts, and latency. This data is essential for tracking costs and usage patterns, as Azure OpenAI charges based on tokens processed. Azure Monitor integrates directly with the service to surface these operational metrics.

Exam trap

The trap here is that candidates confuse high-level cost management (Azure Cost Management) with operational monitoring (Azure Monitor), failing to recognize that Azure Monitor provides the raw token-level data necessary for tracking generative AI usage and costs.

How to eliminate wrong answers

Option A is wrong because Microsoft Purview is a data governance and compliance solution, not a monitoring tool for usage and costs; it focuses on data classification and lineage, not real-time service metrics. Option B is wrong because Azure Advisor provides recommendations for optimizing resource configurations and costs, but it does not offer granular usage or cost tracking for Azure OpenAI Service. Option C is wrong because Azure Cost Management provides high-level cost analysis and budgets across subscriptions, but it lacks the detailed per-request token-level monitoring needed for generative AI usage; it relies on Azure Monitor data for cost breakdowns.

587
MCQhard

You are designing a solution to analyze customer call transcripts using Azure AI Language. The solution must extract key phrases, detect sentiment per utterance, and identify the customer's intent (e.g., 'cancel subscription', 'technical support'). The data is stored in Azure Blob Storage and processed in near real-time. Which combination of Azure AI Language features and processing pattern should you use?

A.Use custom text classification to classify each utterance into intent categories and use the sentiment analysis API on the entire transcript.
B.Use the prebuilt key phrase extraction API to identify important terms and the prebuilt sentiment analysis API for overall transcript sentiment, then map intents via a rules-based approach.
C.Use the conversation summarization API (with utterance-level sentiment and key phrase extraction) and an orchestration workflow model that routes to a custom conversational language understanding project for intent detection.
D.Use the prebuilt conversational language understanding model for intent detection and Azure AI Language sentiment analysis API for utterance-level sentiment, processing each utterance independently via Azure Functions.
AnswerC

Conversation summarization provides utterance-level sentiment and key phrases; orchestration workflow allows routing to a custom CLU project for intent detection, handling multi-turn conversations effectively.

Why this answer

Option D is correct because the conversation summarization API supports utterance-level sentiment and key phrase extraction, and the orchestration workflow model routes intents across different projects. Option A is wrong because the prebuilt conversational language understanding model does not provide sentiment per utterance. Option B is wrong because custom text classification requires separate training and doesn't handle conversation context.

Option C is wrong because the prebuilt key phrase extraction API does not provide sentiment analysis.

588
MCQmedium

A company is implementing a question-answering system using Azure AI Language Service. They have a set of FAQ documents in PDF format. Which feature should they use to automatically generate question-answer pairs?

A.Key Phrase Extraction
B.Extractive Summarization
C.Custom Question Answering
D.Conversational Language Understanding
AnswerC

Custom Question Answering can ingest FAQs and generate Q&A pairs.

Why this answer

The correct answer is A because Custom Question Answering can ingest FAQ documents and generate Q&A pairs. B (Conversational Language Understanding) is for intent recognition. C (Extractive Summarization) summarizes text.

D (Key Phrase Extraction) extracts phrases.

589
MCQhard

You are building a generative AI application that must process large volumes of PDF documents and generate summaries using Azure OpenAI. The solution must be cost-effective and handle variable workloads. Which architecture should you recommend?

A.Use Azure Kubernetes Service (AKS) with a persistent node pool of GPU nodes.
B.Use Azure Functions with a consumption plan to trigger processing jobs and call Azure OpenAI.
C.Deploy a GPU-enabled virtual machine and run the summarization jobs sequentially.
D.Use Azure Logic Apps to iterate through documents and call Azure OpenAI.
AnswerB

Serverless functions scale automatically and you pay only for compute time.

Why this answer

Option B is correct because Azure Functions with a consumption plan provides a serverless, event-driven architecture that scales automatically to handle variable workloads, ensuring cost-effectiveness by charging only for compute time used. This architecture is ideal for processing large volumes of PDFs, as each document can trigger a function execution that calls Azure OpenAI for summarization, without the need for always-on infrastructure.

Exam trap

Microsoft often tests the misconception that GPU or specialized compute is required for AI workloads, but in this scenario, the heavy lifting is done by Azure OpenAI's API, so the focus should be on cost-effective, scalable compute for orchestration, not local GPU processing.

How to eliminate wrong answers

Option A is wrong because Azure Kubernetes Service (AKS) with a persistent node pool of GPU nodes incurs continuous costs even during idle periods, making it less cost-effective for variable workloads, and the GPU nodes are unnecessary since Azure OpenAI is called via API, not run locally. Option C is wrong because deploying a GPU-enabled virtual machine and running summarization jobs sequentially introduces a single point of failure, lacks auto-scaling, and wastes resources on GPU hardware that is not required for API calls. Option D is wrong because Azure Logic Apps is designed for workflow orchestration and integration, not for high-throughput, cost-effective batch processing of large document volumes, and it would incur higher costs per execution compared to Azure Functions.

590
MCQeasy

You are planning a solution that uses Azure AI Language to analyze customer feedback from social media posts. The solution must: - Detect sentiment (positive, negative, neutral) for each post. - Extract key phrases. - Support English and Spanish languages. - Run asynchronously for a batch of 10,000 posts. - Use the least expensive option that meets requirements. What should you do?

A.Use the Azure AI Language service with the built-in sentiment analysis and key phrase extraction capabilities. Process posts in batches using the async API.
B.Build a custom text classification model in Azure AI Language to detect sentiment and extract key phrases.
C.Use the Azure AI Language service with the single-document API for each post.
D.Use Azure AI Translator to translate all posts to English, then use Azure AI Language for analysis.
AnswerA

Built-in features meet requirements at lowest cost; async API handles volume.

Why this answer

Option A is correct because Azure AI Language's built-in sentiment analysis and key phrase extraction natively support both English and Spanish, and the async batch API is designed for high-volume processing (e.g., 10,000 posts) at a lower cost than per-document calls. This approach meets all requirements without custom models or translation overhead.

Exam trap

The trap here is that candidates often assume custom models are required for multilingual support or that translation is necessary, when in fact Azure AI Language's built-in capabilities already cover English and Spanish natively.

How to eliminate wrong answers

Option B is wrong because building a custom text classification model is unnecessary and more expensive; the built-in capabilities already handle sentiment and key phrase extraction for the required languages. Option C is wrong because using the single-document API for each of 10,000 posts would incur higher costs and slower performance compared to the async batch API, which is designed for bulk processing. Option D is wrong because translating all posts to English adds unnecessary cost and latency, and Azure AI Language already supports Spanish natively for both sentiment analysis and key phrase extraction.

591
MCQmedium

You are designing a document processing pipeline using Azure AI Document Intelligence. The pipeline must extract data from both structured forms and unstructured invoices. The solution should support custom models with minimal manual labeling. Which approach should you use?

A.Use prebuilt invoice models for all documents
B.Use Azure AI Language to analyze text
C.Train a custom neural model with a few labeled samples
D.Use OCR and custom regex patterns
AnswerC

Neural models require minimal labeling and handle both structured and unstructured documents.

Why this answer

Option C is correct because Azure AI Document Intelligence's custom neural models can be trained with as few as five labeled samples, leveraging transfer learning to achieve high accuracy on both structured forms and unstructured invoices. This approach minimizes manual labeling effort while supporting the diverse document types in the pipeline, unlike prebuilt models that are limited to specific layouts or rule-based methods that require extensive manual configuration.

Exam trap

The trap here is that candidates assume prebuilt models (Option A) are sufficient for all document types, overlooking the requirement for custom structured forms, or they overestimate the effectiveness of OCR and regex (Option D) without recognizing the need for AI-based layout understanding.

How to eliminate wrong answers

Option A is wrong because prebuilt invoice models are designed for standard invoice layouts and cannot handle custom structured forms or unstructured documents with varying formats, leading to poor extraction accuracy. Option B is wrong because Azure AI Language is a text analytics service for sentiment, key phrases, and entity recognition, not a document extraction service capable of processing structured forms or invoices with field-level extraction. Option D is wrong because OCR with custom regex patterns requires manual definition of patterns for each field, is brittle to layout variations, and does not leverage AI to learn from examples, resulting in high maintenance and poor generalization.

592
MCQmedium

A company is deploying a generative AI solution using Azure OpenAI Service to generate product descriptions. The solution must comply with responsible AI principles, specifically ensuring that generated content does not include harmful or offensive language. Which Azure AI service feature should they implement to automatically filter the output?

A.Enable the default content filtering system in Azure OpenAI Service.
B.Configure Prompt Shields in Azure AI Content Safety.
C.Use Azure AI Content Safety with a custom category severity threshold.
D.Use groundedness detection in Azure AI Content Safety.
AnswerA

Azure OpenAI Service includes a default content filtering system that automatically filters harmful content based on severity levels.

Why this answer

Option A is correct because Azure OpenAI Service includes a built-in default content filtering system that automatically screens generated outputs for harmful or offensive language, aligning with responsible AI principles. This system operates at the service level without additional configuration, making it the simplest and most direct way to filter product descriptions for compliance.

Exam trap

The trap here is that candidates may confuse the built-in content filtering of Azure OpenAI Service with the separate Azure AI Content Safety service, which requires additional setup and is not automatically applied to OpenAI outputs.

How to eliminate wrong answers

Option B is wrong because Prompt Shields in Azure AI Content Safety are designed to protect against prompt injection attacks, not to filter generated content for harmful language. Option C is wrong because Azure AI Content Safety with a custom category severity threshold requires explicit configuration and integration, whereas the question asks for an automatic filtering feature that is already part of Azure OpenAI Service. Option D is wrong because groundedness detection in Azure AI Content Safety checks whether generated content is factually based on source documents, not whether it contains harmful or offensive language.

593
MCQmedium

You are building a solution to extract key information from invoices using Azure AI Document Intelligence. The invoices contain fields such as invoice number, date, total amount, and line items. However, the model is not correctly extracting the line items. Which prebuilt model should you use?

A.Prebuilt-receipt model
B.Prebuilt-idDocument model
C.Prebuilt-invoice model
D.Prebuilt-layout model
AnswerC

Prebuilt-invoice is designed for invoices and extracts line items, totals, and other fields.

Why this answer

Option B is correct because the prebuilt-invoice model is specifically designed to extract fields from invoices, including line items. Option A is wrong because the prebuilt-layout model extracts text and structure but not specific key-value pairs like line items. Option C is wrong because the prebuilt-receipt model is for receipts, not invoices.

Option D is wrong because the prebuilt-idDocument model is for identity documents.

594
MCQhard

A company uses Azure AI Language to analyze customer call transcripts. They need to identify specific entities such as product names and issue types. The prebuilt entity recognition does not cover their custom entities. Which approach should they take to extract both standard and custom entities from the transcripts?

A.Use the prebuilt entity recognition API only and map standard entities to custom categories.
B.Use a single conversational language understanding (CLU) project with entities defined for both standard and custom entities.
C.Use a custom text classification model to classify the transcript and then extract entities from the classified output.
D.Use a custom named entity recognition (NER) model for custom entities and call the prebuilt entity recognition API separately for standard entities.
AnswerD

Combines both approaches to cover all entity types.

Why this answer

Option B is correct because using a custom NER model allows extraction of custom entities, and it can be combined with prebuilt entity extraction via the API. Option A is wrong because a single CLU project is for intent and entity extraction but may not support both prebuilt and custom entities seamlessly. Option C is wrong because using only prebuilt entities misses custom entities.

Option D is wrong because a custom text classification model classifies text but does not extract entities.

595
MCQmedium

A healthcare organization uses Custom Vision to classify X-ray images. They have a small dataset of 200 images per class. Which strategy will most likely improve model accuracy?

A.Reduce the image dimensions to speed up training.
B.Add more negative samples to the dataset.
C.Use data augmentation and transfer learning with a pre-trained model.
D.Increase the number of training iterations significantly.
AnswerC

Data augmentation increases effective dataset size, and transfer learning leverages pre-trained features.

Why this answer

Option A is correct because data augmentation and transfer learning are standard techniques to improve accuracy with small datasets. Option B is wrong because increasing iterations without more data can lead to overfitting. Option C is wrong because reducing image size may lose important details.

Option D is wrong because negative tagging is useful but does not replace the need for more training data.

596
MCQmedium

You deploy the above ARM template to create an Azure AI Vision resource. After deployment, you try to call the Vision API from your application but receive a 403 Forbidden error. What is the most likely cause?

A.The network ACLs block all traffic by default
B.The apiVersion in the template is incorrect
C.The custom subdomain name is not configured correctly
D.The SKU 'S1' does not support API calls
AnswerA

With defaultAction set to 'Deny', no traffic is allowed unless IP rules are added.

Why this answer

The correct answer is that network ACLs are set to Deny by default, blocking all traffic. The template sets networkAcls.defaultAction to 'Deny', meaning no IP addresses are allowed unless explicitly added. Option A is wrong because the SKU S1 is valid.

Option B is wrong because the custom subdomain is correctly configured. Option D is wrong because the apiVersion is valid.

597
MCQmedium

A company is building a solution to analyze customer reviews images using Azure AI Vision. They need to extract text from images that may contain both printed and handwritten text. Which feature should they use?

A.Custom Vision
B.OCR API (optical character recognition)
C.Read API
D.Azure AI Document Intelligence
AnswerC

The Read API is designed to extract both printed and handwritten text from images.

Why this answer

Option B is correct because the Read API is designed to extract printed and handwritten text from images. Option A is wrong because OCR API is older and less capable for handwritten text. Option C is wrong because Form Recognizer is for forms and documents.

Option D is wrong because Custom Vision is for image classification and object detection.

598
MCQhard

You are the AI architect for a global e-commerce company. The company uses Azure AI services to power its product recommendation engine. The current solution uses Azure AI Language to extract product attributes from descriptions and Azure AI Search to index and retrieve products. The company is expanding to new markets and must comply with data residency regulations in the European Union and Asia. Additionally, the solution must handle a 10x increase in traffic during holiday sales without downtime. You need to design a solution that meets these requirements. The solution must use Azure AI Language and Azure AI Search. You have the following options: A) Deploy separate Azure AI Language and Azure AI Search resources in each region (EU and Asia) and use Azure Traffic Manager to route users to the nearest region. Use geo-replication for Azure AI Search. B) Deploy a single Azure AI Language resource in the US and use Azure AI Search with geo-replication to handle traffic. Use Azure Front Door for global routing. C) Deploy Azure AI Language resources in each region and use a global Azure AI Search resource with replication. Use Azure Load Balancer to distribute traffic. D) Deploy Azure AI Language and Azure AI Search resources in the US only, and use Azure CDN to cache responses globally. Which option should you choose?

A.Deploy a single US resource with geo-replication and Front Door.
B.Deploy separate resources in each region with Traffic Manager and geo-replication.
C.Deploy US-only resources with CDN caching.
D.Deploy Language in each region with a global Search resource.
AnswerB

Complies with data residency and handles traffic.

Why this answer

Option B is correct because deploying separate Azure AI Language and Azure AI Search resources in each region (EU and Asia) ensures compliance with data residency regulations by keeping data within regional boundaries. Azure Traffic Manager routes users to the nearest region for low latency, and geo-replication for Azure AI Search provides high availability and disaster recovery, handling the 10x traffic increase during holiday sales without downtime.

Exam trap

The trap here is that candidates may assume a single global resource with geo-replication or caching (Options A, C, D) can satisfy data residency, but they overlook that data processing and storage must both occur within the specific region to comply with regulations, not just be replicated or cached elsewhere.

How to eliminate wrong answers

Option A is wrong because deploying a single Azure AI Language resource in the US violates data residency regulations for EU and Asia, as data would be processed and stored outside those regions. Option C is wrong because using a global Azure AI Search resource with replication still stores data in a single region, failing data residency compliance; Azure Load Balancer does not provide geo-routing or regional isolation. Option D is wrong because deploying Azure AI Language resources in each region but using a global Azure AI Search resource in the US means search data is stored in the US, violating EU and Asia data residency requirements.

599
Multi-Selecthard

Which TWO actions should you take to ensure high availability for an Azure AI service deployed in a single region?

Select 2 answers
A.Configure autoscaling for the AI service.
B.Use a read-only replica of the AI service.
C.Deploy the AI service in a second Azure region.
D.Enable Azure DDoS Protection.
E.Enable Azure Backup for the AI resource.
AnswersA, C

Autoscaling adjusts capacity to maintain performance under load.

Why this answer

Option A is correct because autoscaling allows the Azure AI service to dynamically adjust the number of instances based on demand, ensuring that the service can handle traffic spikes without downtime. This is a key high-availability feature within a single region, as it prevents resource exhaustion and maintains responsiveness. Option C is correct because deploying the AI service in a second Azure region provides geographic redundancy, enabling failover if the primary region experiences an outage, which is a fundamental high-availability strategy.

Exam trap

The trap here is that candidates often confuse high-availability features like autoscaling and multi-region deployment with unrelated services like backup or DDoS protection, or incorrectly assume that read-only replicas apply to AI services as they do to databases.

600
MCQmedium

You are deploying an Azure AI Language service custom text classification model. You need to ensure that the training data is balanced and representative. What should you do?

A.Use only the most frequent labels to train the model.
B.Oversample the minority classes to match the majority class size.
C.Split the labeled data into training and test sets, ensuring each class has similar proportions.
D.Use all labeled data for training and rely on cross-validation.
AnswerC

A ensures balanced representation and proper evaluation.

Why this answer

Option C is correct because splitting labeled data into training and test sets while ensuring each class has similar proportions (stratified split) is a standard practice for balanced and representative training data. This approach prevents class imbalance from skewing model evaluation metrics and ensures the model generalizes well to unseen data. In Azure AI Language custom text classification, this is critical for achieving reliable performance across all classes.

Exam trap

The trap here is that candidates often confuse data balancing techniques (like oversampling) with the fundamental requirement of a representative train-test split, leading them to choose Option B or D instead of recognizing that stratified splitting is the direct and correct method for ensuring balanced and representative data in Azure AI Language custom text classification.

How to eliminate wrong answers

Option A is wrong because using only the most frequent labels discards minority class data entirely, leading to a model that cannot classify underrepresented categories and suffers from severe class imbalance. Option B is wrong because oversampling minority classes to match the majority class size can introduce synthetic duplicates or bias, potentially causing overfitting and not addressing the root need for a representative split; Azure AI Language does not natively support oversampling as a preprocessing step for custom text classification. Option D is wrong because using all labeled data for training without a separate test set prevents proper evaluation of model generalization; cross-validation alone does not guarantee balanced class representation across folds unless explicitly stratified, and Azure AI Language custom training requires a dedicated test set for validation.

Page 7

Page 8 of 14

Page 9