Microsoft Azure AI Engineer Associate AI-102 (AI-102) — Questions 301375

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

Page 4

Page 5 of 14

Page 6
301
MCQeasy

A developer wants to deploy a custom generative AI model using Azure Machine Learning. Which compute target should they choose for low-latency real-time inference?

A.Local deployment
B.Azure Batch
C.Azure Functions
D.Azure Kubernetes Service (AKS)
AnswerD

AKS is designed for real-time inference with low latency.

Why this answer

Azure Kubernetes Service (AKS) is the correct compute target for low-latency real-time inference because it supports horizontal pod autoscaling, GPU acceleration, and can be configured with a low-latency ingress controller (e.g., NGINX or Azure Application Gateway) to route inference requests directly to model containers. AKS also integrates with Azure Machine Learning's real-time inference endpoint, which uses a gRPC or HTTP-based scoring protocol to achieve sub-100ms response times.

Exam trap

The trap here is that candidates often confuse Azure Functions' serverless convenience with real-time capability, overlooking the cold-start penalty and lack of GPU support, while AKS is the only option that provides the necessary infrastructure for consistent low-latency inference.

How to eliminate wrong answers

Option A is wrong because local deployment (e.g., a local Docker container or Jupyter notebook) is intended for development and testing only, not for production-grade low-latency real-time inference, as it lacks scalability, load balancing, and network-level optimizations. Option B is wrong because Azure Batch is designed for high-throughput, parallel batch processing jobs (e.g., offline scoring of large datasets) and is not optimized for low-latency real-time inference due to its job-queue scheduling overhead and lack of persistent endpoints. Option C is wrong because Azure Functions, while serverless and capable of handling HTTP triggers, has a cold-start latency problem (often 1-10 seconds) and limited GPU support, making it unsuitable for sub-second real-time inference workloads.

302
Multi-Selecthard

Which TWO factors should you consider when choosing between Azure AI Document Intelligence and Azure AI Language for extracting information from documents?

Select 2 answers
A.Use of REST APIs
B.Need to process handwritten text
C.Need to extract tables from scanned PDFs
D.Ability to extract key-value pairs
E.Real-time processing requirements
AnswersB, C

Document Intelligence supports handwriting recognition.

Why this answer

Option B is correct because Azure AI Document Intelligence (formerly Form Recognizer) is specifically designed to handle handwritten text through its 'Read' OCR model, which can extract printed and handwritten text from documents. Azure AI Language, on the other hand, focuses on natural language processing (NLP) tasks like sentiment analysis and entity recognition, and does not natively process handwritten content. Therefore, if your document contains handwritten notes, Document Intelligence is the appropriate service.

Exam trap

The trap here is that candidates often assume Azure AI Language can handle all text extraction tasks because of its name, overlooking that Document Intelligence is the specialized service for OCR, layout analysis, and structured data extraction from documents.

303
Multi-Selecthard

You are designing a knowledge mining solution for a large enterprise that uses Azure AI Search to index millions of documents. The solution must support high-availability and automatic failover. Which THREE actions should you take to meet these requirements?

Select 3 answers
A.Use geo-redundant storage (GRS) for the index data.
B.Enable semantic search on the index.
C.Provision the Azure AI Search service in at least two regions.
D.Configure the search service with at least two replicas.
E.Enable indexing of large documents using the text split skill.
AnswersA, C, D

Geo-redundant storage ensures data durability and failover across regions.

Why this answer

Option A ensures redundancy in case of a region failure. Option C enables high availability within a region. Option D ensures data redundancy and failover.

Option B is about query language, not availability. Option E is for performance, not failover.

304
MCQmedium

Your team is building a mobile app that uses Azure AI Computer Vision to extract text from business cards. The app must handle cards in multiple languages (English, French, German). Which feature of the Computer Vision API should you use?

A.OCR API
B.Recognize Text
C.Read API
D.Describe Image
AnswerC

The Read API supports multiple languages and is the recommended option.

Why this answer

Option D is correct because the Read API supports multiple languages and is designed for OCR. The OCR API (A) is older and less accurate. Describe Image (B) generates captions, not text extraction.

Recognize Text (C) is deprecated.

305
Matchingmedium

Match each Azure AI feature to the service that provides it.

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

Concepts
Matches

Text Analytics

Face API

Custom Speech

LUIS

Computer Vision

Why these pairings

These features are part of specific Azure Cognitive Services.

306
Multi-Selecteasy

Which TWO built-in cognitive skills in Azure AI Search can be used to extract entities from text?

Select 2 answers
A.Custom Entity Lookup
B.PII Detection
C.Sentiment Analysis
D.Language Detection
E.Entity Recognition
AnswersB, E

Extracts PII entities.

Why this answer

Options A and C are correct. Entity Recognition extracts named entities (Person, Organization, Location). PII Detection extracts personally identifiable information entities (e.g., phone numbers, emails).

Option B is incorrect because Sentiment Analysis determines sentiment. Option D is incorrect because Language Detection detects language. Option E is incorrect because Custom Entity Lookup matches against a custom list.

307
MCQhard

A company is developing a generative AI solution that must process sensitive customer data. They need to ensure that data remains within their Azure tenant and is not used to improve the base model. Which configuration is required in Azure OpenAI?

A.Opt out of abuse monitoring and data logging in the Azure OpenAI Studio.
B.Enable customer-managed keys for encryption.
C.Use a private endpoint to restrict access to the service.
D.Configure data residency to keep data in a specific region.
AnswerA

Opting out prevents Microsoft from using your data for model improvement.

Why this answer

Option A is correct because opting out of abuse monitoring and data logging in Azure OpenAI Studio ensures that sensitive customer data is not sent to Microsoft for review or used to improve the base model. This configuration is specifically designed for customers who need to process sensitive data while maintaining data residency within their Azure tenant, as it disables the default logging and human review processes that could expose data outside the tenant.

Exam trap

The trap here is that candidates confuse network-level security controls (private endpoints) or encryption (CMK) with data governance policies that control how Microsoft uses data for model training, leading them to select options that address access or storage but not the specific requirement to prevent data from being used to improve the base model.

How to eliminate wrong answers

Option B is wrong because customer-managed keys (CMK) only control encryption at rest and do not prevent data from being used for model improvement or abuse monitoring; CMK is about key ownership, not data governance for training. Option C is wrong because a private endpoint restricts network access to the service but does not affect how Microsoft processes or stores data for model improvement; it only secures the connection path, not the data usage policy. Option D is wrong because data residency configuration ensures data is stored in a specific geographic region but does not prevent Microsoft from using that data for abuse monitoring or base model training; residency is about location, not usage rights.

308
MCQmedium

Your organization has a large set of PDF invoices stored in Azure Blob Storage. You need to extract line-item details (product names, quantities, prices) and store them in Azure SQL Database for downstream reporting. The invoices have varied layouts. Which Azure AI service should you use?

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

Document Intelligence can extract structured data from invoices with varied layouts using prebuilt invoice models.

Why this answer

Option B is correct because Document Intelligence (formerly Form Recognizer) is designed to extract structured data from documents with varied layouts using prebuilt or custom models. Option A is wrong because Computer Vision can extract text but not structured line items. Option C is wrong because Language Service is for NLP tasks, not document extraction.

Option D is wrong because Cognitive Search is for indexing and search, not extraction from documents.

309
MCQhard

A company uses Azure AI Custom Vision to classify images of products. The model is deployed to a mobile app. The app sends images to the Custom Vision prediction endpoint. Users report that the app is slow when the network is poor. You need to enable offline inference on the mobile device. What should you do?

A.Use the Custom Vision compact model with smaller image size
B.Export the model as TensorFlow Lite and integrate it into the mobile app
C.Increase the prediction API timeout
D.Deploy the model to an Azure IoT Edge device
AnswerB

Enables on-device inference without network.

Why this answer

Option D is correct because exporting the model to TensorFlow Lite and running inference on-device allows offline operation. Option A is wrong because it only reduces latency when online. Option B is wrong because it does not enable offline inference.

Option C is wrong because Azure IoT Edge is for edge devices, not mobile apps directly.

310
MCQmedium

You are designing an Azure AI solution that uses Azure AI Document Intelligence to extract data from invoices. The solution must handle up to 1000 invoices per hour with a maximum latency of 5 seconds per invoice. You need to choose the appropriate pricing tier and resource configuration. What should you do?

A.Create a Cognitive Services multi-service account to increase throughput
B.Use the S0 tier and configure autoscaling if needed
C.Use the Free tier and implement a retry policy
D.Deploy the Document Intelligence resource in multiple regions
AnswerB

S0 tier supports higher throughput and meets latency requirements.

Why this answer

The S0 tier (Standard) for Azure AI Document Intelligence supports up to 15 transactions per second (TPS), which equates to 54,000 invoices per hour—far exceeding the 1,000 per hour requirement. With a maximum latency of 5 seconds per invoice, the S0 tier's default throughput is sufficient without needing autoscaling, though autoscaling can be configured for burst scenarios. The Free tier (F0) is limited to 20 transactions per minute (1,200 per hour) and is not designed for production workloads, making S0 the correct choice.

Exam trap

The trap here is that candidates often assume higher throughput requires multi-region deployment or multi-service accounts, but the S0 tier's default TPS limit already exceeds the requirement, making autoscaling optional rather than necessary.

How to eliminate wrong answers

Option A is wrong because creating a Cognitive Services multi-service account does not increase throughput for Document Intelligence; it simply combines multiple AI services under a single endpoint and key, and each service still has its own tier limits. Option C is wrong because the Free tier (F0) is capped at 20 transactions per minute (1,200 per hour) and has a maximum of 5 TPS, which cannot reliably handle 1,000 invoices per hour with 5-second latency, and retry policies do not overcome throughput limits. Option D is wrong because deploying Document Intelligence resources in multiple regions does not increase throughput for a single workload; it is used for geo-redundancy or data residency, not for scaling capacity within a single region.

311
MCQhard

Refer to the exhibit. You are reviewing an Azure OpenAI Service API request. The deployment-id is 'gpt-4o'. The user asks 'What is the capital of France?' The response is cut off mid-sentence. Based on the parameters, what is the most likely cause?

A.The max_tokens value is too low
B.The temperature setting is too high
C.The stop parameter is causing early termination
D.The system message is missing a context
AnswerC

The stop sequence '\n' stops generation prematurely.

Why this answer

Option C is correct because the `stop` parameter in an Azure OpenAI API request defines a sequence of tokens that, when generated, causes the model to stop producing further output. If the `stop` sequence appears in the generated text, the response will be truncated at that point, even mid-sentence. In this scenario, the cut-off response is most likely due to the `stop` parameter matching a token in the generated output, not because of token limits or temperature settings.

Exam trap

Microsoft often tests the distinction between `max_tokens` (which limits total output length) and the `stop` parameter (which causes early termination based on content), leading candidates to mistakenly attribute mid-sentence cut-offs to token limits rather than stop sequences.

How to eliminate wrong answers

Option A is wrong because a low `max_tokens` value would cause the response to stop at the token limit, but the response would not necessarily be cut off mid-sentence; it would simply end after the specified number of tokens, which could be at a natural break. Option B is wrong because a high `temperature` setting increases randomness and creativity but does not cause early termination; it affects the probability distribution of token selection, not the stopping condition. Option D is wrong because the system message provides context or instructions for the model's behavior, but its absence would not cause a mid-sentence cut-off; it might lead to less relevant or less structured responses, but the response would still complete naturally unless another parameter stops it.

312
MCQeasy

You need to extract product codes (e.g., 'PRD-12345') from scanned invoices using Azure AI Document Intelligence. The product codes always follow a pattern of three uppercase letters, a hyphen, and five digits. Which approach should you use?

A.Use the pre-built invoice model in Azure AI Document Intelligence with a regex field extraction
B.Build a custom skill in Azure AI Search using a Python regex
C.Train a custom NER model in Azure AI Language
D.Use Azure OpenAI GPT-4 with document vision to extract the codes
AnswerA

Document Intelligence supports regex-based field extraction in custom models.

Why this answer

Option C is correct because Azure AI Document Intelligence's pre-built invoice model can be extended with custom fields using regex to extract product codes. Option A is wrong because Azure AI Language's custom NER does not handle document layouts. Option B is wrong because Azure AI Search's custom skills require additional development.

Option D is wrong because Azure OpenAI GPT-4 is overkill and less reliable for structured extraction.

313
Multi-Selecthard

Which THREE factors should you consider when choosing between a pre-built model and a custom model in Azure AI Language?

Select 3 answers
A.Domain-specific vocabulary coverage
B.Time to develop and deploy
C.Need for a trained endpoint
D.Availability of labeled training data
E.Model size and memory footprint
AnswersA, B, D

Pre-built may miss domain terms; custom can include them.

Why this answer

Pre-built models are faster to deploy and require no labeled data, but may not cover domain-specific terms. Custom models require labeled data and training time but offer higher accuracy for niche domains. Model size is not a typical consideration; both can be scaled.

Custom models require a trained endpoint. Pre-built models are often cheaper for generic use.

314
MCQhard

You are building a chat bot that uses Azure AI Language to process customer support tickets. The bot must extract entities like order numbers (e.g., ORD-12345) and issue categories. You need to choose the best approach for entity extraction to minimize development effort and ensure high accuracy.

A.Leverage the Text Analytics for Health API to extract entities from the support tickets.
B.Create a custom named entity recognition (NER) project in Azure AI Language that includes prebuilt components for order numbers and trains custom models for categories.
C.Use the Prebuilt Entity Extraction skill in Azure AI Search to extract order numbers and categories from the text.
D.Use the Conversational Language Understanding (CLU) project type to train a model for both intent and entity extraction.
AnswerB

Custom NER with prebuilt components combines ease of use with flexibility for custom categories.

Why this answer

Option B is correct because the prebuilt entity components in the custom NER project can recognize common patterns like order numbers without needing to write custom regex or rules. Option A is wrong because the built-in Prebuilt Entity Extraction skill in Azure AI Search is for indexing, not for the Language service. Option C is wrong because the Conversational Language Understanding (CLU) project type is for intent classification and entity extraction, but it requires a lot of data and training.

Option D is wrong because the Text Analytics for Health is for medical entities and would not work for order numbers.

315
MCQeasy

You need to analyze videos stored in Azure Blob Storage to detect objects and generate timestamps. Which Azure service should you use?

A.Azure Custom Vision
B.Azure Form Recognizer
C.Azure Computer Vision
D.Azure Video Indexer
AnswerD

Video analysis with object detection and timestamps.

Why this answer

Azure Video Indexer (D) is the correct choice because it is specifically designed to analyze videos, extracting insights such as object detection, scene segmentation, and timestamps. It uses AI models to process video content stored in Azure Blob Storage and generates a timeline of detected objects, making it ideal for this scenario.

Exam trap

The trap here is that candidates often confuse Azure Computer Vision (image analysis) with video analysis, overlooking that Computer Vision lacks native video processing and timestamp generation, while Video Indexer is the dedicated service for end-to-end video insights.

How to eliminate wrong answers

Option A is wrong because Azure Custom Vision is a service for training custom image classification and object detection models on images, not for analyzing pre-recorded videos with timestamp generation. Option B is wrong because Azure Form Recognizer is designed to extract text and structure from documents (e.g., invoices, forms), not for video analysis or object detection. Option C is wrong because Azure Computer Vision provides image analysis APIs (e.g., object detection in static images) but lacks native video processing capabilities and timestamp generation; it would require additional custom logic to handle video frames sequentially.

316
MCQmedium

Refer to the exhibit. You are configuring an Azure AI Search skillset. The skillset includes an EntityRecognitionSkill and a KeyPhraseExtractionSkill. After running the indexer, you notice that the 'organizations' field is empty in the index. What is the most likely cause?

A.The skill output path is incorrect
B.The output field mapping for organizations is missing
C.The 'Organization' category is misspelled
D.The skills must be in reverse order
AnswerB

Without mapping the skill output to the index field, the data will not appear.

Why this answer

Option B is correct because the entity recognition skill outputs entities into a specific structure; to get organizations, you need to map the '/document/organizations' path to the index field. Option A is incorrect because the skill is correctly configured to extract organizations. Option C is incorrect because the order of skills doesn't affect entity recognition.

Option D is incorrect because the skill output path is standard.

317
MCQmedium

You are designing a solution that extracts information from scanned invoices. The solution must automatically classify invoices by vendor and extract key fields (total amount, date, invoice number). Which combination of Azure AI services should you use?

A.Azure AI Document Intelligence custom classification and extraction models
B.Azure AI Vision OCR and Azure AI Language
C.Azure AI Form Recognizer (deprecated) with custom models
D.Azure AI Document Intelligence pre-built invoice model
AnswerA

Custom models allow classification by vendor and extraction of specific fields tailored to each vendor's invoice layout.

Why this answer

The correct answer is to use Azure AI Document Intelligence with a custom classification model for vendor classification and a custom extraction model for field extraction. Option A is wrong because pre-built invoice model does not support custom vendor classification. Option B is wrong because Azure AI Vision OCR alone cannot extract structured fields.

Option D is wrong because Form Recognizer is the older name and the service is now Document Intelligence; the combination is correct but the naming is outdated, and it does not include classification.

318
Multi-Selecteasy

You need to use Azure AI Language to analyze customer feedback. Which THREE analysis types are available in the Text Analytics API?

Select 3 answers
A.Image captioning
B.Speech-to-text
C.Sentiment analysis
D.Entity recognition
E.Key phrase extraction
AnswersC, D, E

Analyzes sentiment.

Why this answer

Options A, B, and C are correct because these are standard features. Option D is wrong because image analysis is part of Azure AI Vision. Option E is wrong because speech-to-text is part of Azure AI Speech.

319
MCQhard

Your company uses Azure AI Search to power a customer-facing product catalog search. The search index contains product data from Azure SQL Database. The indexer runs daily. Lately, users complain that new products appear in the catalog with a delay of up to 24 hours. The business requires near real-time indexing (within minutes) for new products. You have the following constraints: - The indexer must continue to run daily for full sync. - You need to minimize changes to the existing architecture. - You cannot use Azure Functions or Logic Apps due to cost. What should you do?

A.Add a high-water mark change detection policy (e.g., based on a LastModified column) to the indexer data source. Set the indexer to run every 5 minutes.
B.Enable change tracking on the Azure SQL Database and configure the indexer to use it.
C.Modify the application to use the push API to upload new products to the index as they are added.
D.Remove the daily indexer schedule and instead trigger the indexer manually after each new product insertion.
AnswerA

This enables incremental indexing with minimal changes, achieving near real-time updates.

Why this answer

Option D is correct. By creating a data change detection policy (e.g., using a LastModified timestamp), the indexer can track incremental changes. Setting the indexer schedule to run every 5 minutes allows near real-time updates.

The daily full sync can still be run separately. Option A is wrong because the push API requires code changes and additional infrastructure (e.g., a service to push changes). Option B is wrong because Azure SQL Database change tracking is more complex and not necessary.

Option C is wrong because removing the schedule would require manual runs, not solving the delay.

320
MCQmedium

A retail company uses Azure AI Vision to analyze in-store video feeds. They want to detect when employees are not wearing required safety vests. The solution must send real-time alerts to store managers. Video feeds are captured from existing CCTV cameras. Which approach should you recommend?

A.Use Azure AI Vision Image Analysis with a custom model to analyze each frame offline
B.Train a Custom Vision classification model and deploy it to the edge on CCTV cameras
C.Use Azure AI Video Indexer with a custom object detection model to detect vests in video frames
D.Use Azure Video Analyzer for Media to index videos and detect vests after recording
AnswerC

Supports real-time video and custom model deployment.

Why this answer

Azure AI Video Indexer supports real-time video analysis with custom models. Azure AI Vision Image Analysis works on still images, not real-time video. Custom Vision is for image classification.

Azure Video Analyzer for Media (now part of Video Indexer) is correct.

321
MCQhard

You run the Azure CLI command shown in the exhibit to create an online endpoint for a generative AI model. The deployment fails because the selected VM instance type is not available in the East US region. Which action should you take to resolve the issue?

A.Increase the instance count to 2
B.Specify a different VM type or region that supports Standard_NC6s_v3
C.Use a batch endpoint instead of online endpoint
D.Change --compute-type to CPU
AnswerB

Choosing an available VM type or region resolves the issue.

Why this answer

The deployment failed because the Standard_NC6s_v3 VM instance type is not available in the East US region. The correct action is to either choose a different VM type that is available in East US or deploy to a different region that supports Standard_NC6s_v3. This directly addresses the root cause of the failure, as Azure Machine Learning online endpoints require the selected VM SKU to be available in the target region.

Exam trap

The trap here is that candidates may think increasing instance count or switching to a batch endpoint will bypass the regional SKU limitation, but neither changes the underlying VM type or region, so the deployment will still fail.

How to eliminate wrong answers

Option A is wrong because increasing the instance count does not change the VM type or region; it only scales out the number of instances, which does not resolve the unavailability of the VM SKU. Option C is wrong because switching to a batch endpoint does not address the VM availability issue; batch endpoints also require compatible compute resources and are designed for asynchronous, large-scale inference, not for fixing regional SKU unavailability. Option D is wrong because changing --compute-type to CPU would not help if the model requires GPU acceleration (as implied by the NC-series VM), and it does not solve the regional availability problem for the specified VM type.

322
MCQeasy

You need to extract printed text from scanned invoices in multiple languages. Which Azure service should you use?

A.Azure Form Recognizer
B.Azure Computer Vision Read API
C.Azure Custom Vision
D.Azure Video Indexer
AnswerB

OCR service supporting multiple languages.

Why this answer

Option A is correct because Azure Computer Vision's Read API is designed for OCR with multilingual support. Option B is wrong because Custom Vision is for custom image classification, not OCR. Option C is wrong because Form Recognizer is for form extraction, but the Read API is the primary OCR service.

Option D is wrong because Video Indexer is for video.

323
MCQmedium

Refer to the exhibit. You deploy an Azure AI Services multi-service account using this ARM template. After deployment, developers cannot access the service from their local machines. What is the most likely reason?

A.The location is not specified correctly
B.The pricing tier S0 does not support network restrictions
C.The custom subdomain name is not globally unique
D.The network ACLs block all traffic by default
AnswerD

'defaultAction': 'Deny' blocks all traffic unless explicitly allowed via IP rules or virtual networks.

Why this answer

The ARM template for an Azure AI Services multi-service account includes a `networkAcls` property with `defaultAction` set to `Deny`. This configuration blocks all traffic by default, including requests from developers' local machines, unless an explicit IP rule or virtual network rule is added to allow access. Since no such rules are shown in the exhibit, the network ACLs are the most likely cause of the connectivity failure.

Exam trap

Microsoft often tests the misconception that network ACLs only affect virtual network traffic or that they are optional by default, when in fact setting `defaultAction: Deny` explicitly blocks all traffic unless allow rules are configured.

How to eliminate wrong answers

Option A is wrong because the location is specified as `eastus` in the ARM template, which is a valid Azure region and does not affect network access control. Option B is wrong because the S0 pricing tier does support network restrictions; network ACLs are available for all tiers of Azure AI Services multi-service accounts, including S0. Option C is wrong because the custom subdomain name `myaiservices` is used only for endpoint resolution and does not impact network-level access; global uniqueness is required for subdomain names but a conflict would cause a deployment failure, not a post-deployment access issue.

324
MCQmedium

You are developing a conversational agent using Microsoft Copilot Studio that must handle complex multi-turn conversations. The agent needs to maintain context across multiple user inputs. Which feature should you use?

A.Use actions to call external APIs for context.
B.Use topics with slots to collect information across turns.
C.Use global variables to store conversation state.
D.Use entities to capture user input.
AnswerB

Topics with slots manage multi-turn context.

Why this answer

Option B is correct because Copilot Studio uses topics with slots to manage context and gather information across turns. Option A is wrong because entities alone don't manage conversation flow. Option C is wrong because variables are used to store data but slots are for context.

Option D is wrong because actions are for API calls, not conversation management.

325
MCQeasy

You are using Azure AI Language to perform entity recognition on customer feedback. You need to identify the sentiment expressed towards specific entities. Which feature should you use?

A.Named Entity Recognition (NER)
B.Sentiment analysis with opinion mining
C.Entity linking
D.Key phrase extraction
AnswerB

Opinion mining provides sentiment at the entity or aspect level.

Why this answer

Option C is correct because sentiment analysis with opinion mining provides sentiment at the aspect level, including entities. Option A is wrong because Named Entity Recognition (NER) only identifies entities, not sentiment. Option B is wrong because key phrase extraction only extracts key phrases.

Option D is wrong because entity linking connects entities to knowledge bases, not sentiment analysis.

326
MCQhard

You are building a generative AI application using Azure OpenAI Service. The application must provide factual answers based on your company's internal knowledge base. You need to minimize the risk of the model generating incorrect information (hallucinations). Which approach should you take?

A.Implement Retrieval-Augmented Generation (RAG) with Azure AI Search.
B.Fine-tune the model on your company's documents.
C.Use few-shot prompting with examples of correct answers.
D.Set the max_tokens parameter to a low value.
AnswerA

RAG retrieves relevant documents and uses them as context, reducing hallucinations.

Why this answer

Option A is correct because Retrieval-Augmented Generation (RAG) with Azure AI Search grounds the model's responses in your company's internal knowledge base by retrieving relevant documents in real time and injecting them into the prompt. This reduces hallucinations by ensuring the model generates answers based on retrieved facts rather than relying solely on its parametric memory. Azure AI Search provides vector and hybrid search capabilities that efficiently index and query your documents, making RAG the most effective approach for factual accuracy.

Exam trap

Cisco often tests the misconception that fine-tuning (Option B) is the best way to ground a model in proprietary data, but the trap is that fine-tuning does not provide dynamic, query-specific retrieval and can still produce hallucinations, whereas RAG explicitly forces the model to use retrieved facts.

How to eliminate wrong answers

Option B is wrong because fine-tuning adjusts the model's weights on your documents, which can lead to overfitting and does not guarantee that the model will not hallucinate; it still relies on its internal knowledge and may generate plausible-sounding but incorrect information when queried outside the fine-tuned distribution. Option C is wrong because few-shot prompting provides examples but does not ground the model in your specific knowledge base; the model can still hallucinate if the examples do not cover the exact query context or if it extrapolates incorrectly. Option D is wrong because setting max_tokens to a low value only truncates the output length and does not improve factual accuracy; it may even cause incomplete or misleading answers without addressing the root cause of hallucinations.

327
MCQmedium

A company is deploying a solution using Azure AI Vision to analyze images of products on a retail website. They need to ensure that the image analysis is performed within a specific geographic boundary for data residency compliance. What should they configure?

A.Deploy the Azure AI Vision resource in the desired Azure region
B.Create a private endpoint for the Vision resource
C.Enable multi-region replication on the Vision resource
D.Use the Free tier of Azure AI Vision
AnswerA

Data stays in the region where the resource is deployed.

Why this answer

Azure AI Vision resources are regional Azure resources, meaning all data processing and storage occur within the Azure region where the resource is deployed. By deploying the resource in the desired geographic region, you ensure that image analysis and any derived data remain within that boundary, satisfying data residency compliance requirements. This is the fundamental mechanism for controlling data location in Azure AI services.

Exam trap

The trap here is that candidates confuse network isolation (private endpoints) with data residency, or assume that replication or tier changes can alter where data is stored, when in fact the region of the resource itself is the sole determinant for compliance.

How to eliminate wrong answers

Option B is wrong because a private endpoint restricts network access to the resource via a private IP address in your virtual network, but it does not control the geographic location where data is processed or stored. Option C is wrong because Azure AI Vision does not support multi-region replication; that feature is available for Azure Storage and Cosmos DB, not for AI services. Option D is wrong because the Free tier imposes usage limits (e.g., 20 transactions per minute) and does not provide any data residency guarantees; it still processes data in the region where the resource is created.

328
MCQhard

You are analyzing an image using the Azure AI Vision REST API with the JSON request above. The response includes a description: 'a person holding a smartphone'. However, the response does not include any brand information even though the smartphone is clearly visible. What is the most likely reason?

A.The smartphone brand is not in the Microsoft brand catalog.
B.The 'brands' feature was not included in the request.
C.The 'model-version' is set to 'latest' which does not support brand detection.
D.The API only detects one brand per image, and a different brand was detected.
AnswerA

Brand detection only recognizes brands in Microsoft's predefined catalog.

Why this answer

Option D is correct. The brands feature is included, but brand detection only works for well-known brands; if the smartphone brand is not in Microsoft's brand catalog, it won't be detected. Option A is wrong because 'brands' is included.

Option B is wrong because model-version 'latest' supports brand detection. Option C is wrong because the API can detect multiple brands.

329
Drag & Dropmedium

Drag and drop the steps to create a custom Azure AI Translator 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

Start with parallel data, create the resource, upload to custom model, train, then deploy.

330
MCQhard

Your Azure AI Search index is experiencing high query latency. You have enabled semantic search and custom scoring profiles. You need to reduce latency without degrading search quality. Which action should you take?

A.Remove custom scoring profiles.
B.Increase the number of replicas.
C.Reduce the number of partitions.
D.Disable semantic search.
AnswerB

More replicas distribute query load and reduce latency.

Why this answer

Increasing the number of replicas distributes query load across multiple copies of the index, allowing parallel processing of search requests. This directly reduces query latency without altering the search logic, scoring profiles, or semantic enrichment, thus preserving search quality.

Exam trap

The trap here is that candidates confuse partitions (which affect storage and indexing speed) with replicas (which affect query throughput), leading them to incorrectly reduce partitions or disable features instead of scaling query capacity.

How to eliminate wrong answers

Option A is wrong because removing custom scoring profiles would degrade search quality by eliminating relevance tuning, and it does not address the root cause of high latency (insufficient query capacity). Option C is wrong because reducing partitions decreases the index's data capacity and can increase latency by forcing more data per partition, while partitions primarily affect indexing speed and storage, not query throughput. Option D is wrong because disabling semantic search would degrade search quality by removing AI-powered ranking and relevance features, and it does not address the underlying query load issue that replicas solve.

331
MCQmedium

You are developing a chat application that uses Azure OpenAI Service to answer customer queries. You need to ensure that the model does not generate responses containing internal company policies or confidential information. Which approach should you use?

A.Use Azure AI Search with index filtering to exclude documents with confidential info.
B.Fine-tune the model on a dataset that excludes confidential information.
C.Set the system message to instruct the model not to include confidential information.
D.Configure Azure AI Content Safety with custom categories for confidential terms.
AnswerC

System messages guide model behavior effectively for content restrictions.

Why this answer

Option B is correct because system messages provide instructions to the model on how to behave, including restricting certain topics. Option A is wrong because Azure AI Content Safety filters harmful content but does not prevent model from generating confidential info. Option C is wrong because fine-tuning may not guarantee avoidance of specific topics.

Option D is wrong because embeddings are for retrieval, not for restricting model output.

332
Multi-Selectmedium

You are using Azure OpenAI Service to generate text. You need to reduce the likelihood of the model generating repetitive sequences. Which TWO parameters should you adjust?

Select 2 answers
A.frequency_penalty
B.max_tokens
C.top_p
D.temperature
E.presence_penalty
AnswersA, E

Frequency penalty reduces repetition by penalizing frequent tokens.

Why this answer

Frequency penalty reduces the likelihood of the model repeating the same tokens or phrases by subtracting a fixed penalty from the log-probability of tokens that have already appeared in the generated text. This directly discourages repetitive sequences, making it one of the two correct parameters to adjust.

Exam trap

The trap here is that candidates often confuse temperature and top_p with repetition control, but these parameters affect randomness and diversity of token selection, not the direct penalization of repeated tokens.

333
MCQeasy

You are developing a solution that uses Azure AI Language to analyze customer feedback. You need to determine whether the sentiment of a given sentence is positive, negative, or neutral. Which Azure AI Language feature should you use?

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

Sentiment Analysis directly provides sentiment labels and scores.

Why this answer

Option A is correct because Sentiment Analysis is the feature that returns sentiment labels and scores. Option B is wrong because Key Phrase Extraction identifies key phrases but not sentiment. Option C is wrong because Language Detection identifies the language.

Option D is wrong because Entity Recognition identifies entities.

334
MCQmedium

A company is building a knowledge mining solution using Azure AI Search. They need to extract key phrases from a large set of documents in multiple languages. Which skill should they add to the skillset?

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

Key Phrase Extraction is designed to extract key phrases from text.

Why this answer

The Key Phrase Extraction skill is the correct choice because it is specifically designed to identify and extract the most important phrases from text, which directly supports the requirement to extract key phrases from documents. Azure AI Search's built-in Key Phrase Extraction skill leverages natural language processing to analyze text and return a list of key phrases, making it the appropriate skill for this knowledge mining solution.

Exam trap

The trap here is that candidates may confuse Entity Recognition (which extracts single-word entities like 'Microsoft') with Key Phrase Extraction (which extracts multi-word phrases like 'Azure AI Search'), leading them to choose Option D instead of A.

How to eliminate wrong answers

Option B (Sentiment Analysis skill) is wrong because it evaluates the emotional tone or sentiment (positive, negative, neutral) of text, not the extraction of key phrases. Option C (Language Detection skill) is wrong because it identifies the language of the text but does not extract key phrases from the content. Option D (Entity Recognition skill) is wrong because it identifies and categorizes named entities (e.g., people, organizations, locations) rather than extracting multi-word key phrases that summarize the document's main topics.

335
MCQhard

Your company uses Azure Cognitive Search to index millions of documents. Users report that search results include irrelevant documents. You need to improve search relevance by boosting documents that contain the search term in the title field. Which scoring profile configuration should you use?

A.Create a tagging scoring profile that boosts by the title field with a weight of 10.
B.Create a freshness scoring profile with a boosting duration of 30 days.
C.Create a distance scoring profile with a reference point parameter.
D.Create a magnitude scoring profile with a boosting function of 'linear'.
AnswerA

A tagging profile boosts documents that have matching terms in a specific field, like title.

Why this answer

Option A is correct because a tagging profile with a weight boosts documents based on matching tags from a specific field. Option B is wrong because a distance profile is for geospatial scoring. Option C is wrong because a magnitude profile is for boosting based on numeric values.

Option D is wrong because a freshness profile boosts based on date/time.

336
MCQmedium

You manage an Azure AI Search solution that indexes documents from Azure Blob Storage. The index must support real-time updates when documents are added or modified. Which approach should you use?

A.Use the push API to manually upload documents.
B.Configure an indexer with a short schedule and change tracking.
C.Manually reset and rerun the indexer after each change.
D.Add a cognitive skillset to process new documents.
AnswerB

Indexers can detect changes in Blob Storage and update the index automatically.

Why this answer

Option B is correct because Azure AI Search indexers can be configured with a short schedule (e.g., every 5 minutes) and change tracking (using high-water mark or integrated change detection on Azure Blob Storage) to automatically index new or modified documents without manual intervention. This provides near-real-time updates while leveraging the indexer's built-in change detection capabilities.

Exam trap

The trap here is that candidates often confuse the push API (Option A) as the only way to achieve real-time updates, overlooking that indexers with change tracking and a short schedule can provide automated near-real-time indexing without custom code.

How to eliminate wrong answers

Option A is wrong because the push API requires manual coding to upload documents, which does not automatically detect changes in Azure Blob Storage and is not a 'configured' approach for real-time updates from a data source. Option C is wrong because manually resetting and rerunning the indexer after each change is not automated and does not support real-time updates; it is a manual, batch-oriented process. Option D is wrong because a cognitive skillset enriches documents with AI transformations (e.g., OCR, entity recognition) but does not handle the scheduling or change tracking needed for real-time indexing of new or modified documents.

337
MCQmedium

A development team is using Azure Cognitive Service for Language to extract key phrases from customer reviews. They notice that some reviews are not being processed, and the API returns a 400 error code. What is the most likely cause?

A.One of the reviews exceeds the maximum character limit for a single document.
B.The reviews contain characters that are not valid UTF-8.
C.The request contains more than 5 documents.
D.The reviews are written in a language not supported by the service.
AnswerA

The service limits each document to 5,120 characters.

Why this answer

The Azure Cognitive Service for Language key phrase extraction API enforces a maximum document size of 5,120 characters per document. When a single review exceeds this limit, the API returns a 400 Bad Request error because the request payload violates the service's input constraints. This is the most common cause of 400 errors in batch text analysis operations.

Exam trap

The trap here is that candidates often assume the 400 error is due to unsupported languages or encoding issues, but the actual constraint is the per-document character limit, which is explicitly documented in the service's input specifications.

How to eliminate wrong answers

Option B is wrong because the service automatically handles UTF-8 encoding validation and would return a different error (e.g., 400 with 'InvalidRequestContent') if characters were not valid UTF-8, but the question states the reviews are standard customer reviews, making invalid UTF-8 unlikely. Option C is wrong because the API supports up to 10 documents per request (not 5), so a request with more than 5 documents would still succeed unless it exceeds the 10-document limit. Option D is wrong because unsupported languages typically result in a successful response with empty key phrases or a warning, not a 400 error; the service supports over 40 languages for key phrase extraction.

338
MCQmedium

You are developing a custom text classification model using Azure AI Language. The model must classify customer support tickets into 15 categories. You have 10,000 labeled examples. After training, the model shows 95% accuracy on the test set but only 60% on a small sample of new tickets. What is the most likely cause?

A.The model's confidence threshold is set too low.
B.The training data is not representative of the new tickets.
C.There is data leakage between the training and test sets.
D.The model is overfitting to the training data.
AnswerD

Overfitting leads to high training accuracy but poor generalization.

Why this answer

Option C is correct because high accuracy on the test set but low accuracy on new data indicates overfitting, where the model learns the training data too well and fails to generalize. Option A is wrong because accuracy on the test set is high, so insufficient training data is not indicated. Option B is wrong because low confidence scores are not mentioned; overfitting can still produce high confidence.

Option D is wrong because data leakage would cause inflated accuracy on the test set, but the test set accuracy is already high and the problem is with new data.

339
MCQhard

You are developing a chatbot for a retail company using Azure AI Language's custom question answering. The chatbot must provide answers from a knowledge base of 500 FAQ documents. Users often ask the same question in different wording, and the chatbot fails to return an answer for paraphrased queries. What is the most effective solution?

A.Use Azure AI Bot Service's Direct Line Speech channel to improve accuracy.
B.Enable active learning in the project settings and periodically publish the updated knowledge base.
C.Increase the number of FAQ documents in the knowledge base.
D.Manually add alternate question phrases to the knowledge base for each QnA pair.
AnswerB

Active learning automatically suggests alternative phrasings based on user queries.

Why this answer

Option C is correct because enabling active learning allows the system to learn from user interactions and suggest alternative phrasings. Option A is wrong because increasing document count doesn't help with paraphrasing. Option B is wrong because additional training is not a built-in feature; active learning is.

Option D is wrong because Azure AI Bot Service is for bot hosting, not improving answer coverage.

340
MCQeasy

You are building an Azure AI Search solution to index a collection of technical manuals. Users need to find documents by searching for specific terms and also have the ability to filter by document category. Which feature should you configure in the index to support filtering?

A.Set the 'filterable' property to true on the category field
B.Set the 'facetable' property to true on the category field
C.Set the 'searchable' property to true on the category field
D.Set the 'sortable' property to true on the category field
AnswerA

Filterable fields allow OData filter expressions to be applied in queries.

Why this answer

Option A is correct because filterable fields in Azure AI Search allow filtering of search results. Option B is wrong because facetable enables drill-down navigation, not direct filtering. Option C is wrong because sortable does not enable filtering.

Option D is wrong because searchable determines full-text search, not filtering.

341
Multi-Selectmedium

You are building a custom question answering solution using Azure AI Language. Which TWO actions are required to deploy the solution?

Select 2 answers
A.Create a Custom Question Answering project in Azure AI Language.
B.Train a custom model using the Azure AI Language training API.
C.Create a QnA Maker service in the Azure portal.
D.Deploy the project to an Azure AI Language resource.
E.Publish the project to a web app bot.
AnswersA, D

This is the first step.

Why this answer

Options B and D are correct because you need to create a project and deploy the model for production use. Option A is wrong because the QnA Maker service is deprecated; Azure AI Language uses the Custom Question Answering feature. Option C is wrong because training data can be provided in multiple formats and the model is trained automatically.

Option E is wrong because publishing to a web app bot is optional.

342
MCQmedium

Refer to the exhibit. You are creating a Conversational Language Understanding (CLU) project using the Azure AI Language Service REST API. You want the project to support both English and Spanish utterances. Which parameter in the request body enables this?

A.projectName
B.multilingual
C.language
D.confidenceThreshold
AnswerB

Setting multilingual to true enables support for multiple languages.

Why this answer

The correct answer is D because the 'multilingual' parameter set to true enables support for multiple languages. A (confidenceThreshold) is for prediction confidence. B (projectName) is just the name.

C (language) sets the primary language but does not enable multilingual support.

343
MCQeasy

Your organization needs to monitor Azure AI services for unusual activity patterns that might indicate a security threat. Which Microsoft security solution should you use?

A.Microsoft Intune
B.Microsoft Defender XDR
C.Microsoft Purview
D.Microsoft Sentinel
AnswerD

Sentinel provides SIEM and SOAR capabilities for security monitoring.

Why this answer

Microsoft Sentinel is a cloud-native Security Information and Event Management (SIEM) and Security Orchestration Automated Response (SOAR) solution. It is specifically designed to ingest logs from Azure AI services, apply analytics to detect unusual activity patterns, and generate alerts for potential security threats, making it the correct choice for monitoring AI services for security anomalies.

Exam trap

The trap here is that candidates often confuse Microsoft Sentinel with Microsoft Defender XDR, assuming that 'security monitoring' always falls under Defender, but Sentinel is the SIEM solution required for ingesting and analyzing logs from Azure AI services, while Defender XDR focuses on endpoint and identity protection.

How to eliminate wrong answers

Option A is wrong because Microsoft Intune is a mobile device management (MDM) and mobile application management (MAM) solution, focused on managing endpoints and enforcing compliance policies, not on monitoring cloud service activity for security threats. Option B is wrong because Microsoft Defender XDR (Extended Detection and Response) is designed to correlate signals across endpoints, email, and identities, but it does not natively ingest and analyze logs from Azure AI services for SIEM-style threat detection. Option C is wrong because Microsoft Purview is a data governance and compliance solution, primarily used for data cataloging, classification, and policy enforcement, not for real-time security monitoring or threat detection.

344
Multi-Selecthard

Which THREE factors should be considered when choosing between Azure AI Vision prebuilt models and Custom Vision?

Select 3 answers
A.Custom Vision cannot process images from real-time video feeds.
B.Prebuilt models require a large training dataset.
C.Prebuilt models may not have high accuracy for industry-specific objects.
D.Prebuilt models are suitable for common object categories like cars and animals.
E.Custom Vision allows training on custom object categories.
AnswersC, D, E

Prebuilt models are trained on general data, so accuracy may be lower for niche objects.

Why this answer

Options A, C, and E are correct. Prebuilt models have limited accuracy for niche domains (A). Custom Vision allows training on specific objects (C).

Prebuilt models cover common objects, but Custom Vision is needed for custom categories (E). Option B is wrong because Custom Vision also supports images from various sources. Option D is wrong because prebuilt models do not require training data.

345
MCQhard

Refer to the exhibit. You receive this error when calling an Azure Cognitive Services API. What is the most likely cause?

A.The API endpoint is incorrect.
B.The subscription key has expired.
C.The subscription key is from a different region.
D.The subscription key is not valid for this Cognitive Services resource.
AnswerD

Inner error explicitly states that.

Why this answer

The error clearly states invalid subscription key. The inner error confirms it's not valid for the resource.

346
MCQeasy

A company wants to use Azure AI Translator to translate customer emails from English to French. They need to ensure that the translation preserves the tone and formality of the original text. What should they configure in the request?

A.Set the 'category' parameter to 'general' to use a standard translation model.
B.Set the 'scope' parameter to 'document' to ensure context-aware translation.
C.Set the 'formality' parameter to the desired level (e.g., 'formal' or 'informal').
D.Set the 'language' parameter to 'fr' and the 'from' parameter to 'en'.
AnswerC

The formality parameter controls the tone of the translation.

Why this answer

Option C is correct because Azure AI Translator provides a 'formality' parameter that allows you to specify the desired level of formality (e.g., 'formal' or 'informal') in the translated text. This parameter directly controls the tone and register of the output, ensuring that the translation preserves the original email's tone and formality, which is critical for customer communications.

Exam trap

The trap here is that candidates often confuse the 'formality' parameter with language or category settings, mistakenly thinking that simply specifying the target language (Option D) or using a general category (Option A) is sufficient to control tone, when in fact the formality parameter is the only dedicated mechanism for this purpose.

How to eliminate wrong answers

Option A is wrong because the 'category' parameter is used to select a custom translation model or domain (e.g., 'general' for standard translations), but it does not control tone or formality; it affects terminology and style based on the training domain. Option B is wrong because Azure AI Translator does not have a 'scope' parameter; context-aware translation for documents is handled by the Document Translation feature, not by a request parameter in the standard Translate operation. Option D is wrong because while setting 'language' to 'fr' and 'from' to 'en' is necessary for specifying the source and target languages, it does not address the requirement to preserve tone and formality; it only defines the language pair.

347
MCQeasy

You need to summarize a large document using Azure AI Language. Which feature should you use?

A.Document summarization
B.Key phrase extraction
C.Entity recognition
D.Sentiment analysis
AnswerA

Document summarization provides a concise summary of the document.

Why this answer

Option B is correct because Document summarization (extractive or abstractive) is designed to summarize documents. Option A is wrong because Key phrase extraction extracts phrases, not summaries. Option C is wrong because Entity recognition identifies entities.

Option D is wrong because Sentiment analysis gives sentiment scores.

348
MCQhard

You are designing a solution to detect brand logos in social media images. The logos vary in size and orientation. You need to achieve high accuracy with minimal false positives. Which approach should you recommend?

A.Use Azure Computer Vision Describe API to generate captions and filter by logo mentions.
B.Train an Azure Custom Vision object detection model with labeled logo images.
C.Use Azure Computer Vision Analyze API with domain-specific models.
D.Use Azure Form Recognizer to extract logo positions from images.
AnswerB

Custom object detection can learn to detect logos in various conditions.

Why this answer

Option B is correct because Custom Vision with object detection can be trained on diverse logo images to handle variations. Option A is wrong because the Analyze API does not detect specific objects. Option C is wrong because Form Recognizer is for documents.

Option D is wrong because the Describe API generates captions, not logo detection.

349
MCQhard

You are reviewing the skillset definition for an Azure AI Search indexer. The SplitSkill splits the document content into pages of 5000 characters. The SentimentSkill is set to run on each page. However, the sentiment analysis is not producing correct results. What is the most likely cause?

A.The maximumPageLength of 5000 is too high for sentiment analysis
B.The input source for SentimentSkill should be '/document/pages/*' but the SplitSkill output is named 'pages', so the input should be '/document/pages'
C.The context of the SentimentSkill is set to an array, which is not supported
D.The SentimentSkill uses an incorrect @odata.type version
AnswerB

The input source should map to the output of the split skill, which is named 'pages'.

Why this answer

Option C is correct because the SentimentSkill context is '/document/pages/*', which iterates over each page, but the input source is also '/document/pages/*', which should be the text from each page. However, the issue is that the SentimentSkill expects a single string, but the SplitSkill outputs an array. The input should be '/document/pages/*' but that is an array; the correct mapping would be to use the split output correctly.

Actually, the error is that the input source should be '/document/pages/*' which is the array item, but the sentiment skill might be receiving an array instead of a string if not properly configured. Option C identifies that the input source should be the textItems from the split output. Option A is wrong because the skill version is fine.

Option B is wrong because context can be arrays. Option D is wrong because the maximumPageLength is valid.

350
MCQmedium

You are testing an Azure OpenAI chat completion. The response shown in the exhibit is returned. What does the finish_reason of 'content_filter' indicate?

A.The model's response was blocked by the content filter.
B.There was a system error during processing.
C.The user's prompt was flagged by the content filter.
D.The model refused to answer due to insufficient data.
AnswerA

The finish_reason indicates the response was filtered.

Why this answer

The 'content_filter' finish_reason indicates that the Azure OpenAI content filtering system detected that the model's generated response violated one of the configured content policies (e.g., hate, violence, self-harm, sexual content). The response was therefore blocked before being returned to the user, and the finish_reason explicitly signals this filtering action rather than a normal completion or a stop due to token limits.

Exam trap

The trap here is that candidates often confuse 'content_filter' with prompt rejection, but the finish_reason specifically indicates the model's output was blocked, not the user's input.

How to eliminate wrong answers

Option B is wrong because a system error during processing would return a different finish_reason such as 'error' or an HTTP 500 status, not 'content_filter'. Option C is wrong because the content_filter finish_reason applies to the model's response, not the user's prompt; if the prompt were flagged, the API would typically return a 400 error with a content filter violation message before any generation occurs. Option D is wrong because the model refusing to answer due to insufficient data would be indicated by a finish_reason of 'stop' (if the model generated a refusal message) or by a specific response text, not by the 'content_filter' reason.

351
Matchingmedium

Match each Azure Cognitive Search skill to its capability.

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

Concepts
Matches

Extract text from images

Identify entities like people or organizations

Extract key phrases from text

Detect language of text

Determine sentiment of text

Why these pairings

These are built-in cognitive skills in Azure Cognitive Search.

352
Drag & Dropmedium

Drag and drop the steps to deploy a custom language model using Azure AI Language 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, have labeled data ready, then create the resource, train the model, evaluate, and deploy.

353
Multi-Selecthard

You are architecting an Azure AI solution that uses Azure AI Language to analyze text for sentiment and key phrases. The solution must handle bursts of up to 500 requests per second but average only 50 requests per second. You need to ensure cost efficiency while meeting performance requirements. Which THREE actions should you take?

Select 3 answers
A.Use Azure Queue Storage to buffer requests during spikes
B.Select the Free tier and implement queuing
C.Use the S0 pricing tier
D.Implement client-side throttling and retry logic
E.Deploy the service in multiple regions
AnswersA, C, D

Decouples ingestion and processing, smoothing out bursts.

Why this answer

Option A is correct because Azure Queue Storage can decouple the ingestion of requests from processing, allowing the solution to buffer bursts of up to 500 requests per second while the Azure AI Language service processes at a lower sustained rate. This prevents request throttling and enables cost-efficient scaling by using a lower-cost queue to absorb spikes rather than over-provisioning the AI service tier.

Exam trap

The trap here is that candidates often assume a higher pricing tier (like S0) alone can handle bursts, but without queuing and retry logic, the service will still throttle requests, leading to failures or the need for costly over-provisioning.

354
MCQmedium

A company is building an agent that uses Azure OpenAI Service to answer customer queries by querying a SQL database. The agent must be able to handle complex multi-turn conversations and maintain context. Which approach should the team use to implement the agent?

A.Use a single prompt that includes the entire conversation history and the latest user question, then generate SQL.
B.Use a chain-of-thought prompt to generate SQL queries directly from user input, without maintaining conversation history.
C.Use embeddings-based retrieval to find relevant past interactions and include them in the prompt.
D.Use a conversational agent framework like AutoGen with a tool that executes SQL queries, and maintain conversation state.
AnswerD

AutoGen provides multi-turn conversation management and tool integration.

Why this answer

Option D is correct because AutoGen is a conversational agent framework designed for multi-turn, stateful interactions. It can maintain conversation context across turns and integrate a tool to execute SQL queries, which directly meets the requirement for complex multi-turn conversations with context retention. The other options either lack state management or rely on stateless prompt engineering, which is insufficient for maintaining context in a multi-turn agent.

Exam trap

Microsoft often tests the distinction between stateless prompt engineering (options A, B, C) and stateful agent frameworks (option D), where candidates mistakenly believe that simply including history in a prompt (option A) is sufficient for multi-turn context, ignoring token limits and the lack of structured state management.

How to eliminate wrong answers

Option A is wrong because using a single prompt with the entire conversation history quickly exceeds the token limit of the Azure OpenAI model (e.g., 4096 or 8192 tokens for GPT-4), leading to truncation or loss of context, and it does not provide a structured mechanism for maintaining conversation state across turns. Option B is wrong because chain-of-thought prompting without maintaining conversation history cannot handle multi-turn conversations; each turn would be treated as an isolated query, losing all prior context and making it impossible to handle follow-up questions or references to previous interactions. Option C is wrong because embeddings-based retrieval finds semantically similar past interactions but does not inherently maintain the current conversation's state or context; it is typically used for retrieval-augmented generation (RAG) in single-turn scenarios, not for managing the sequential state of a multi-turn dialogue.

355
MCQhard

Your company is deploying an Azure AI solution that uses multiple AI services. You need to ensure that all API calls are authenticated securely using managed identities. Which of the following steps is required to enable managed identity authentication for an Azure AI service?

A.Enable system-assigned managed identity at the subscription level
B.Assign a managed identity to the Azure AI service and grant it the required RBAC role
C.Store the authentication key in Azure Key Vault and reference it in the application
D.Configure a shared access key in the Azure AI service
AnswerB

This enables secure authentication without secrets.

Why this answer

Option B is correct because managed identity authentication for Azure AI services requires assigning either a system-assigned or user-assigned managed identity to the resource, and then granting that identity the appropriate RBAC role (e.g., Cognitive Services User) on the target AI service. This eliminates the need for keys or secrets in the code and leverages Azure AD tokens for secure, passwordless authentication.

Exam trap

The trap here is that candidates often confuse managed identity with key management solutions like Key Vault, or assume that enabling a managed identity at a higher scope (subscription) automatically applies to all resources, when in fact the identity must be explicitly assigned to each resource and granted RBAC permissions.

How to eliminate wrong answers

Option A is wrong because managed identities are assigned at the resource level, not the subscription level; enabling a system-assigned identity at the subscription scope is not a valid operation. Option C is wrong because storing the authentication key in Key Vault is a valid security practice but does not use managed identity authentication—it still relies on a static key, not an Azure AD token. Option D is wrong because shared access keys are the traditional key-based authentication method, which managed identities are designed to replace; configuring a shared access key does not enable managed identity authentication.

356
MCQhard

You are a solution architect at a financial services company. You need to implement a knowledge mining solution that extracts information from annual reports (PDF) of publicly traded companies. The reports contain financial tables, executive summaries, and legal disclaimers. The solution must: (1) extract the company name, fiscal year, revenue, net income, and CEO name; (2) redact any personally identifiable information (PII) like email addresses and phone numbers before indexing; (3) index the extracted data in Azure AI Search; (4) allow users to query using natural language (e.g., 'Which company had the highest revenue in 2023?'). The reports are uploaded to an Azure Blob Storage container. You have access to Azure AI Services and Azure OpenAI. Which combination of services and configurations should you use?

A.Use Azure AI Document Intelligence custom extraction model trained on annual reports to extract fields. In the Azure AI Search pipeline, add a PII detection skill to redact PII. Enable semantic search for natural language queries.
B.Use Azure AI Vision OCR to extract text from PDFs, then use Azure AI Language to extract entities and key phrases. Index in Azure AI Search with semantic search.
C.Use Azure AI Search with blob indexer, include a skillset with Document Layout skill, Entity Recognition skill (for financial entities), and Key Phrase Extraction. Enable semantic search.
D.Use Azure OpenAI GPT-4 to process each report via a custom extraction prompt, then send extracted JSON to Azure AI Search. Enable semantic search.
AnswerA

Best approach for structured extraction, PII redaction, and natural language query.

Why this answer

Option C is correct because it uses Document Intelligence for structured extraction (tables, financial data), PII detection skill for redaction, and semantic search for natural language queries. Option A lacks PII redaction. Option B uses OpenAI for extraction but may be less reliable for structured tables and lacks PII redaction.

Option D uses OCR but Document Intelligence is better for digital PDFs.

357
MCQhard

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

A.The SKU S0 does not support API calls from applications
B.The resource kind is set incorrectly for Language Service
C.The network ACLs block all traffic because no IP rules are defined
D.The custom subdomain name is invalid
AnswerC

Deny default with no allowed IPs blocks all calls.

Why this answer

The correct answer is A because the template sets 'defaultAction' to 'Deny' with no IP rules, blocking all network access. B (S0 SKU) supports API calls. C (custom subdomain) does not cause 403.

D (kind) is correct for Language Service.

358
MCQmedium

You run the Azure CLI command 'az search indexer list --search-service mysearch --query "[].{name:name, status:status, lastResult:lastResult}"' and get the above output. Your indexer shows 5 warnings. What should you do to investigate the warnings?

A.Run 'az search indexer run --name myindexer' to trigger a new run.
B.Run 'az search indexer show --name myindexer' and review the 'warnings' array in the output.
C.Ignore the warnings because they are not errors.
D.Run 'az search indexer reset --name myindexer' to reset the indexer.
AnswerB

The indexer show command returns detailed execution history including warnings.

Why this answer

Warnings indicate issues that did not cause failure but should be reviewed. Use the 'show' command with --debug or check the indexer execution history via portal or REST API. The reset command would re-run without addressing warnings.

359
MCQhard

Refer to the exhibit. You are configuring a custom conversational language understanding project in Azure AI Language. The project currently has English training phrases. You need to add support for Spanish. The exhibit shows the current settings. What should you do?

A.Create a new project for Spanish
B.Set enableMultiLanguage to true and add Spanish training phrases
C.Set enableMultiLanguage to true and deploy
D.Use Azure AI Translator to translate English phrases to Spanish
AnswerB

Enables multi-language and provides training data.

Why this answer

Option B is correct because in Azure AI Language, a custom conversational language understanding (CLU) project can support multiple languages by setting the `enableMultiLanguage` property to `true` and then adding training phrases in the target language (Spanish) to the same project. This allows the model to learn intents and entities across languages without creating separate projects, leveraging the multilingual capabilities of the underlying LUIS-based engine.

Exam trap

The trap here is that candidates often assume they need separate projects for each language (Option A) or that enabling multilingual mode alone is sufficient without adding training phrases in the target language (Option C), overlooking the requirement to provide actual language-specific examples for the model to learn from.

How to eliminate wrong answers

Option A is wrong because creating a new project for Spanish would duplicate effort and miss the built-in multilingual support in Azure AI Language, which is designed to handle multiple languages within a single project when `enableMultiLanguage` is enabled. Option C is wrong because simply setting `enableMultiLanguage` to `true` and deploying without adding Spanish training phrases would not teach the model Spanish utterances, resulting in no Spanish support. Option D is wrong because Azure AI Translator is a separate service for text translation, not for training a CLU model; translating phrases would not create native Spanish training data that captures cultural or linguistic nuances, and the model would still need Spanish training phrases to learn intents correctly.

360
MCQmedium

Refer to the exhibit. You are deploying an agent in Microsoft Foundry using the ARM template snippet above. The agent needs to call Microsoft Graph API to reset a user's password. However, the deployment fails with an authorization error. What is the most likely cause?

A.The Graph API endpoint URL is incorrect.
B.The apiVersion '2025-01-01-preview' is not supported for agent deployments.
C.The managed identity does not have the 'User.ReadWrite.All' delegated permission for Microsoft Graph.
D.The resourceId for managed identity is missing the 'Microsoft.ManagedIdentity/userAssignedIdentities' resource type.
AnswerC

The identity must be granted Graph API permissions.

Why this answer

Option B is correct because the managed identity needs Graph API permissions assigned. Option A is wrong because the endpoint is valid. Option C is wrong because the resourceId is correctly specified.

Option D is wrong because the apiVersion is a preview version but that doesn't cause authorization errors.

361
MCQeasy

Refer to the exhibit. You have a Custom Question Answering project configured with the JSON shown. When you test the project in Azure AI Language Studio, the query 'How many vacation days do I get?' returns no answer. What is the most likely cause?

A.The query is not phrased as an exact match to the trained question.
B.The language is set to English but the query uses informal language.
C.The answer field is empty in the JSON.
D.The confidence score threshold is set too high.
AnswerA

Custom Question Answering matches questions based on semantic similarity, but if the phrasing is too different, it may not return an answer.

Why this answer

The JSON shows that the question is phrased as 'What is the vacation policy?' and the answer is about accruing 15 days. However, the test query 'How many vacation days do I get?' is semantically different. Custom Question Answering uses a fixed set of question-answer pairs and does not automatically rephrase queries; it relies on matching the intent.

Option C is correct because the query does not match the trained question. Option A is incorrect because the language is set to English. Option B is incorrect because the answer is present.

Option D is incorrect because the confidence threshold is not specified in the exhibit.

362
MCQeasy

You need to provision an Azure AI Language resource that supports custom text classification. Which pricing tier should you select to use the custom training feature?

A.Free (F0)
B.Basic (B)
C.Standard (S)
D.Premium (P)
AnswerC

Supports custom text classification training.

Why this answer

The Standard (S) tier is the only pricing tier for Azure AI Language that supports custom text classification, including the custom training feature. The Free (F0) tier is limited to pre-built capabilities and cannot be used for training custom models, while Basic (B) and Premium (P) tiers do not exist for this service.

Exam trap

The trap here is that candidates may confuse the Azure AI Language pricing tiers with other Azure AI services (e.g., Azure Cognitive Search or Azure Bot Service) that offer Basic or Premium tiers, leading them to select a non-existent option for this specific service.

How to eliminate wrong answers

Option A is wrong because the Free (F0) tier only supports pre-built language features (e.g., sentiment analysis, key phrase extraction) and explicitly disables custom model training and deployment. Option B is wrong because Azure AI Language does not offer a Basic (B) pricing tier; the available tiers are Free (F0) and Standard (S). Option D is wrong because Azure AI Language does not have a Premium (P) tier; the highest tier for custom features is Standard (S).

363
MCQhard

You are designing a responsible AI solution for a healthcare application that uses Azure AI Vision to analyze medical images. The solution must minimize bias across demographic groups and provide explainability for predictions. Which combination of services should you use?

A.Azure AI Document Intelligence for extracting metadata and Azure AI Language for sentiment analysis.
B.Azure AI Translator for multilingual support and Azure AI Metrics Advisor for monitoring.
C.Azure AI Search for indexing results and Azure AI Anomaly Detector for outliers.
D.Azure AI Content Safety for content moderation and Fairlearn with Azure Machine Learning interpretability.
AnswerD

Content Safety filters inappropriate content; Fairlearn and interpretability address bias and explainability.

Why this answer

Option D is correct because Azure AI Content Safety helps filter harmful content and reduce bias in model outputs, while Fairlearn with Azure Machine Learning interpretability provides tools to assess fairness across demographic groups and explain model predictions. This combination directly addresses the requirements of minimizing bias and providing explainability for medical image analysis.

Exam trap

The trap here is that candidates may confuse general-purpose AI services (like translation or search) with specialized fairness and interpretability tools, overlooking that Fairlearn and ML interpretability are the Azure-native solutions for responsible AI requirements.

How to eliminate wrong answers

Option A is wrong because Azure AI Document Intelligence extracts text from documents and Azure AI Language performs sentiment analysis, neither of which addresses bias minimization or explainability for image analysis. Option B is wrong because Azure AI Translator provides multilingual translation and Azure AI Metrics Advisor monitors time-series data, both irrelevant to bias detection or model interpretability in medical imaging. Option C is wrong because Azure AI Search indexes searchable content and Azure AI Anomaly Detector identifies outliers in time-series data, neither of which offers fairness assessment or explainability for AI vision predictions.

364
MCQmedium

A company uses Microsoft Copilot Studio to create an agent that helps employees schedule meetings. The agent must access the user's calendar to find free time slots and book meetings. The agent should only work for users who have granted consent. Which authentication and authorization approach should be used?

A.Configure OAuth 2.0 authentication with Microsoft Entra ID and request delegated permissions for Microsoft Graph.
B.Use certificate-based authentication for the agent.
C.Use API key authentication to call Microsoft Graph.
D.Use OAuth 2.0 client credentials flow with application permissions.
AnswerA

Delegated permissions allow the agent to act as the user, and consent ensures user authorization.

Why this answer

Option A is correct because OAuth 2.0 with delegated permissions allows the agent to act on behalf of the user after consent. Option B is wrong because application permissions are for app-only access, not user delegation. Option C is wrong because API key does not support user context.

Option D is wrong because certificate-based auth is also app-only.

365
Multi-Selectmedium

You are designing a generative AI solution that uses Azure OpenAI Service. The solution must not generate responses that include personally identifiable information (PII). Which TWO configurations should you implement? (Choose two.)

Select 2 answers
A.Use Azure AI Content Safety to filter PII in the output.
B.Enable diagnostic logging to audit all generated responses.
C.Set data retention policies to delete prompts and completions after 30 days.
D.Configure the system message to instruct the model not to generate PII.
E.Fine-tune the model on a dataset that excludes PII.
AnswersA, D

Content Safety can detect and block PII.

Why this answer

A and D are correct. Using system messages to instruct the model to avoid PII is a direct approach. Azure AI Content Safety can detect and filter PII in outputs.

B is wrong because fine-tuning on non-PII data does not guarantee avoidance. C is wrong because data retention policies affect storage, not generation. E is wrong because enabling logging does not prevent generation of PII.

366
MCQhard

You are responsible for an Azure AI solution that uses Custom Vision to classify manufacturing defects. The model must achieve high recall to avoid missing defects. The current model has high precision but low recall. Which action should you take?

A.Add more images of non-defective items to the training set
B.Increase the number of training iterations
C.Lower the probability threshold for the defect class
D.Increase the probability threshold for the defect class
AnswerC

Lowering the threshold increases the number of positive predictions, thus improving recall.

Why this answer

Lowering the probability threshold for the defect class means the model will classify an image as defective even when its confidence score is lower. This increases the number of true positives (defects caught), directly improving recall at the cost of potentially more false positives. In Custom Vision, the default probability threshold is 50%, and adjusting it downward is the standard technique to prioritize recall over precision.

Exam trap

The trap here is that candidates confuse precision and recall, often assuming that increasing the threshold (making the model stricter) will improve overall performance, when in fact it reduces recall by missing more defects.

How to eliminate wrong answers

Option A is wrong because adding more images of non-defective items would bias the model toward the non-defective class, likely reducing recall further by making the model more conservative in predicting defects. Option B is wrong because increasing the number of training iterations (epochs) primarily helps the model converge better on the training data but does not directly control the precision-recall trade-off; it may even lead to overfitting without improving recall. Option D is wrong because increasing the probability threshold for the defect class would require higher confidence to classify a defect, which reduces false positives but also reduces true positives, thereby lowering recall even further.

367
MCQhard

You are using Azure OpenAI Service to generate product descriptions. You notice that the model occasionally outputs descriptions that contain factual inaccuracies about product specifications. You want to reduce these hallucinations without changing the model. What should you do?

A.Increase the frequency_penalty parameter.
B.Decrease the temperature parameter.
C.Increase the max_tokens parameter.
D.Provide the product specifications in the prompt and use the system message to instruct the model to base answers on them.
AnswerD

Grounding the model with factual data in the prompt reduces hallucinations by providing accurate context.

Why this answer

Option D is correct because providing the product specifications directly in the prompt and using the system message to instruct the model to base its answers on them grounds the generation in factual data, reducing hallucinations. This technique, known as 'grounding' or 'retrieval-augmented generation' (RAG), does not modify the model itself but constrains its output to the provided context, which is the only way to reduce factual inaccuracies without changing model parameters.

Exam trap

The trap here is that candidates often confuse hyperparameter tuning (like temperature or frequency_penalty) with prompt engineering techniques, mistakenly believing that adjusting randomness or repetition penalties can fix factual hallucinations, when only providing the correct context in the prompt can do so without model changes.

How to eliminate wrong answers

Option A is wrong because increasing frequency_penalty reduces repetition of tokens by penalizing tokens that have already appeared, which does not address factual accuracy or hallucinations. Option B is wrong because decreasing temperature makes the model more deterministic and less creative, but it does not prevent the model from generating plausible-sounding but factually incorrect statements about product specifications. Option C is wrong because increasing max_tokens only allows longer responses, which can actually increase the chance of hallucinations by giving the model more opportunity to generate unsupported content.

368
MCQhard

An application uses Azure AI Vision to analyze images and extract text. The application crashes when processing images with embedded barcodes. You suspect the issue is related to the image pre-processing. Which step should you add to the pipeline to resolve the issue?

A.Increase the image contrast before sending to the OCR engine
B.Add more training data with barcodes to the OCR model
C.Detect and remove barcodes from the image before OCR
D.Resize the image to a smaller resolution to reduce barcode impact
AnswerC

Removing barcodes eliminates patterns that cause OCR errors, improving accuracy.

Why this answer

The correct answer is to add a barcode detection and removal step before OCR. Barcodes can confuse OCR services because they contain patterns that may be interpreted as text, causing errors. Option A is wrong because adjusting contrast does not address barcode interference.

Option B is wrong because adding more training data is for custom models, not pre-built OCR. Option D is wrong because compressing the image may reduce quality and exacerbate the issue.

369
MCQhard

A company uses the Face API for identity verification. During testing, they find that the similarity scores between two images of the same person are lower than expected. Which factor is most likely causing this?

A.The images are compressed with different quality levels.
B.The images have different lighting conditions (e.g., one is brightly lit, the other is dark).
C.The backgrounds of the images are different.
D.The images have different dimensions (e.g., 500x500 vs 1000x1000).
AnswerB

Lighting changes facial appearance and reduces similarity scores.

Why this answer

Option B is correct because large differences in lighting can significantly affect similarity scores. Option A is wrong because moderate difference in image size does not affect facial features. Option C is wrong because background mismatch does not affect the face comparison.

Option D is wrong because compression artifacts may have a minor effect but not as much as lighting.

370
MCQeasy

Your team is building a mobile app that uses Azure Custom Vision to classify plant species. The app must work offline and sync labeled images when connectivity is restored. Which SDK feature should you use?

A.Azure IoT Edge runtime on the phone
B.Azure API Management with caching
C.Export the model as a TensorFlow or CoreML model for on-device inference
D.Continuous deployment integration
AnswerC

Exported model runs offline.

Why this answer

Option B is correct because Custom Vision provides an offline prediction capability by exporting the model to run locally on device. Option A is wrong because continuous deployment is for cloud deployments. Option C is wrong because API Management is for managing APIs, not offline operation.

Option D is wrong because IoT Edge is for edge devices, not mobile apps.

371
MCQmedium

You are designing an AI solution that uses Azure OpenAI Service to answer customer queries. The solution must ensure that the model does not generate harmful or inappropriate content. Which Azure AI service should you configure to enforce content safety policies?

A.Azure AI Speech
B.Azure OpenAI Service
C.Azure AI Content Safety
D.Azure AI Search
AnswerC

Azure AI Content Safety provides content moderation and filtering.

Why this answer

Azure AI Content Safety is the dedicated service for detecting and filtering harmful content such as hate speech, violence, self-harm, and sexual content. It provides configurable severity thresholds and can be integrated with Azure OpenAI Service to enforce content safety policies on model inputs and outputs. This ensures the AI solution meets responsible AI requirements without modifying the underlying model.

Exam trap

The trap here is that candidates assume Azure OpenAI Service has built-in configurable content safety policies, but in reality, it only provides default safety systems; custom policies require Azure AI Content Safety.

How to eliminate wrong answers

Option A is wrong because Azure AI Speech is focused on speech-to-text, text-to-speech, and speech translation capabilities, not on content moderation or safety policy enforcement. Option B is wrong because Azure OpenAI Service itself does not include built-in content safety policy configuration; it relies on Azure AI Content Safety for such filtering, though it has default safety systems, those are not user-configurable for custom policies. Option D is wrong because Azure AI Search is a cognitive search service for indexing and querying data, not for content moderation or safety filtering.

372
MCQhard

A developer is building an agent using the Microsoft Bot Framework SDK in C#. The agent must authenticate users via Microsoft Entra ID and maintain state across conversations. The solution must store user preferences (e.g., language, timezone) in Azure Cosmos DB. Which state management approach should the developer use?

A.Use the Bot State Service (deprecated).
B.Use UserState with Blob Storage.
C.Use ConversationState with Memory Storage.
D.Use UserState with Cosmos DB Storage.
AnswerD

UserState persists user-specific data across conversations, and Cosmos DB is a scalable storage option.

Why this answer

Option D is correct because UserState with Cosmos DB storage provides a scalable solution for storing user preferences across conversations. Option A is wrong because ConversationState is scoped to a single conversation. Option B is wrong because Blob Storage is for large objects, not key-value state.

Option C is wrong because Bot State Service is deprecated.

373
MCQmedium

Your organization uses Microsoft Entra ID for identity management. You are building an AI solution that uses Azure AI Vision to analyze images. The solution must use managed identities to authenticate to the Vision resource. Which RBAC role should you assign to the managed identity?

A.Owner
B.Cognitive Services User
C.Reader
D.Contributor
AnswerB

This role allows API access without management permissions.

Why this answer

The Cognitive Services User role (B) is the correct RBAC role because it grants the minimum required permissions for a managed identity to call Azure AI Vision APIs (e.g., analyze image, OCR) without allowing any write or management operations. This role is specifically designed for accessing Azure Cognitive Services endpoints, and it aligns with the principle of least privilege for authentication via managed identities.

Exam trap

The trap here is that candidates often confuse the Reader role (which grants read access to the resource's Azure Resource Manager properties) with the ability to read data from the service, but Reader does not include data-plane permissions for Cognitive Services APIs.

How to eliminate wrong answers

Option A (Owner) is wrong because it grants full control over the Vision resource, including the ability to delete or modify the resource itself, which is excessive and violates security best practices for a managed identity that only needs to call APIs. Option C (Reader) is wrong because it only allows read access to the resource's metadata and configuration (e.g., viewing keys or endpoints) but does not grant permission to call the Vision API endpoints for image analysis. Option D (Contributor) is wrong because it allows creating and managing resources (e.g., deploying models or changing settings) but does not include the specific 'Cognitive Services User' data-plane permission required to authenticate and invoke the Vision API.

374
MCQhard

You are designing a solution that uses Azure AI Translator to translate customer support tickets. The solution must handle confidential data and ensure no data leaves the specified Azure region. What should you do?

A.Enable customer-managed keys and disable logging
B.Use the global Translator endpoint
C.Create a Translator resource with a private endpoint and disable public network access
D.Use the S1 pricing tier to ensure regional processing
AnswerC

Keeps data within the region and private network.

Why this answer

Option C is correct because using a private endpoint ensures that all traffic to the Translator resource stays within the Azure virtual network and never traverses the public internet, while disabling public network access enforces that no data can leave the specified Azure region. This combination meets the requirement for handling confidential data with regional data residency guarantees.

Exam trap

The trap here is that candidates often confuse encryption (CMK) or pricing tiers with network-level data residency controls, failing to realize that only private endpoints combined with disabled public access can guarantee data never leaves the specified region.

How to eliminate wrong answers

Option A is wrong because customer-managed keys (CMK) control encryption at rest, not data residency or network isolation, and disabling logging does not prevent data from leaving the region. Option B is wrong because the global Translator endpoint routes traffic through Microsoft's global network and may process data outside the specified Azure region, violating the data residency requirement. Option D is wrong because the S1 pricing tier only affects throughput and capacity, not the geographic processing location; regional processing is determined by the resource's region and network configuration, not the pricing tier.

375
MCQhard

You are deploying an Azure OpenAI model for a healthcare application. You need to ensure that the model does not generate medical advice and that all responses include a disclaimer. Which configuration should you use?

A.Ground the model with your own medical documents.
B.Set max_tokens to 50 to limit response length.
C.Use Azure AI Content Safety to filter medical terms.
D.Configure a system message with instructions and enable content filtering.
AnswerD

System message can instruct disclaimer; content filtering blocks prohibited content.

Why this answer

Option D is correct because configuring a system message with explicit instructions (e.g., 'Do not provide medical advice; always include a disclaimer') combined with Azure AI Content Safety's content filtering allows you to enforce behavioral guardrails and block harmful outputs at the application layer. The system message sets the model's behavior, while content filtering provides a secondary safety net to catch policy violations, ensuring compliance in a regulated healthcare environment.

Exam trap

The trap here is that candidates confuse content filtering with behavioral control, assuming Azure AI Content Safety can enforce custom rules like 'do not generate medical advice' when it only filters predefined harmful categories, not domain-specific instructions.

How to eliminate wrong answers

Option A is wrong because grounding the model with medical documents (e.g., via Azure OpenAI on your data) does not prevent the model from generating medical advice; it only improves factual accuracy by referencing your data, but the model can still produce advice or omit disclaimers. Option B is wrong because setting max_tokens to 50 limits response length but does not control the content or ensure a disclaimer is included; the model could still generate medical advice within that token limit. Option C is wrong because Azure AI Content Safety filters harmful content based on predefined categories (e.g., hate, violence), but it does not have a built-in 'medical terms' filter; it cannot enforce a custom rule like 'do not generate medical advice' or 'include a disclaimer'.

Page 4

Page 5 of 14

Page 6