Courseiva

Microsoft Certified: Azure AI Apps and Agents Developer Associate (AI-103) (AI-103) (AI-103) — Questions 226300

510 questions total · 7pages · All types, answers revealed

Page 3

Page 4 of 7

Page 5
226
Multi-Selecthard

You are designing a disaster recovery and high availability strategy for an enterprise Azure OpenAI deployment. Which THREE architectural components or practices should you include? Each correct answer presents part of the solution.

Select 3 answers
A.Configure Azure Monitor metric alerts to detect latency spikes or availability drops.
B.Implement client-side retry and failover logic to handle transient errors and regional failovers.
C.Rely exclusively on Azure Backup to snapshot live model memory states.
D.Deploy secondary Azure OpenAI resources in a paired Azure region.
E.Store all model weights locally on client laptops for offline execution.
AnswersA, B, D

Alerts enable rapid detection of performance degradation triggering failover.

Why this answer

Multi-region deployments, client-side retry/failover logic, and monitoring metrics are essential for DR.

227
MCQhard

You are implementing cross-domain document extraction where incoming PDFs contain complex nested tables with merged cells. You need to extract these tables accurately into structured JSON. Which capability of Azure AI Document Intelligence should you rely on?

A.The documents collection, which converts tables into Markdown format automatically.
B.The keyValuePairs collection, which automatically flattens table rows into key-value pairs.
C.The styles collection, which identifies table gridlines through computer vision edge detection.
D.The tables collection in the analyze result, which provides rowSpan, colSpan, and cell bounding regions.
AnswerD

The tables object structure provides detailed cell properties including rowSpan and colSpan to handle merged cells.

Why this answer

Document Intelligence layout and custom models provide table extraction capabilities that identify row index, column index, span, and whether cells are headers or merged.

228
Multi-Selectmedium

You are implementing opinion mining in an application that analyzes hotel reviews. Which THREE pieces of information does opinion mining return for an evaluated sentence? (Choose three.)

Select 3 answers
A.Compiler error logs for client application code
B.Sentiment polarity (positive, negative, neutral) for the aspect
C.Target aspect (e.g., 'room', 'staff')
D.Assessment evaluation (e.g., 'clean', 'rude')
E.Network bandwidth utilization metrics
AnswersB, C, D

Polarity is evaluated per aspect target.

Why this answer

Opinion mining returns aspect targets (e.g., 'bed'), assessments (e.g., 'comfortable'), and sentiment polarities (positive, negative, neutral) for each aspect.

229
Multi-Selecthard

Which THREE features are provided by the Azure AI Foundry evaluation service?

Select 3 answers
A.Visualization of evaluation results
B.Built-in metrics for RAG (e.g., coherence, fluency)
C.Automated model fine-tuning based on failures
D.Custom evaluation code integration
E.Automatic user interface generation
AnswersA, B, D

Results are visualized in the portal.

Why this answer

Built-in metrics, custom evaluation logic, and dataset visualization are key features.

230
MCQhard

You need to implement token-based authentication using Azure AD (Microsoft Entra ID) for a Python application consuming an Azure OpenAI resource. Which authorization header format must your application use to authenticate successfully?

A.Authorization: API-Key <api_key_value>
B.X-MS-Token: <access_token>
C.Authorization: Bearer <access_token>
D.Ocp-Apim-Subscription-Key: <access_token>
AnswerC

Microsoft Entra ID tokens require the Bearer scheme in the Authorization header.

Why this answer

When using Microsoft Entra ID authentication for Azure AI services, requests must include a Bearer token obtained from the token endpoint, formatted as Authorization: Bearer <access_token>.

231
MCQmedium

You are developing a generative AI application that uses Azure OpenAI. You want to track token consumption per user department for chargeback purposes. Which feature should you enable and utilize?

A.Azure Cost Management budgeting alerts only
B.Azure AI Content Safety severity reports
C.Azure Advisor security recommendations
D.Azure Monitor metrics and diagnostic logs with custom metadata tags
AnswerD

Diagnostic settings send detailed request logs (including token counts) to Log Analytics where queries can aggregate consumption by tags.

Why this answer

Enabling Azure OpenAI usage metrics combined with custom headers or Microsoft Entra ID claims allows tracking and auditing token consumption by department.

232
Multi-Selectmedium

Which THREE types of tools can be integrated into an Azure AI Agent?

Select 3 answers
A.Azure AI Search indexes.
B.Custom Python functions.
C.Hard drive file systems (local).
D.External REST APIs.
E.Operating system kernel drivers.
AnswersA, B, D

Search indexes are standard RAG tools.

Why this answer

Agents can integrate with search indexes, custom Python functions, and external REST APIs.

233
MCQeasy

You are creating a custom prompt template in Azure AI Foundry. You want to insert the user's input safely while preventing prompt injection vulnerabilities. How should you design the template structure?

A.Remove all spaces and punctuation from the user input
B.Concatenate user input directly into the middle of the system prompt string
C.Use explicit XML tags or markdown delimiters to separate system instructions from user inputs
D.Encode all user inputs using Base64 before sending to the model
AnswerC

Delimiters and XML tags clearly demarcate untrusted user input from system instructions.

Why this answer

Separating system instructions from user input using clear delimiters (such as markdown blockquotes or XML tags like <user_input>) helps the model distinguish instructions from data.

234
MCQmedium

You are analyzing survey responses. You want to categorize each response into one of three predefined business categories ('Billing', 'Technical Support', 'General Inquiry'). Which Azure AI Language feature should you use?

A.Key phrase extraction
B.Custom text classification
C.Extractive summarization
D.Prebuilt Named Entity Recognition
AnswerB

Custom text classification trains a model to classify documents into user-defined categories.

Why this answer

Custom text classification allows you to train a model to categorize documents into your own custom-defined categories.

235
Multi-Selectmedium

You are reviewing the JSON response of a sentiment analysis API call from the Azure AI Language service. Which THREE properties or confidence scores are returned in the standard response payload for a given document?

Select 3 answers
A.Sentence-level sentiment and confidence scores
B.Confidence scores for positive, neutral, and negative sentiment classes
C.Exact translation of the text into English
D.Overall document sentiment (e.g., positive, negative, neutral, mixed)
E.Grammar and spelling correction suggestions
AnswersA, B, D

The response includes a breakdown of sentiment results at the individual sentence level.

Why this answer

The sentiment analysis response structure includes the overall document sentiment, confidence scores for positive, neutral, and negative classes, sentence-level breakdowns, and opinion mining components if enabled.

236
MCQeasy

You are provisioning a multi-service Azure AI resource using an Azure Resource Manager (ARM) template. Which resource type should you specify in the template for a generic multi-service cognitive services account?

A.Microsoft.AI/cognitiveServices
B.Microsoft.CognitiveServices/projects
C.Microsoft.MachineLearningServices/workspaces
D.Microsoft.CognitiveServices/accounts
AnswerD

Microsoft.CognitiveServices/accounts is the correct resource provider type for provisioning AI services in ARM or Bicep templates.

Why this answer

The standard resource type for a multi-service Cognitive Services account in ARM templates is 'Microsoft.CognitiveServices/accounts'.

237
MCQeasy

You are building a customer service agent in Azure AI Foundry. The agent needs to call an external inventory management REST API when customers ask about stock levels. Which feature of Azure AI Agent Service should you implement to allow the agent to interact with this external API?

A.Function calling (Tools)
B.Prompt shields
C.Azure AI Content Safety filters
D.Azure Cognitive Search index
AnswerA

Function calling enables the model to output JSON arguments that your application code uses to invoke external APIs.

Why this answer

Function calling (or tool definitions) allows agents to invoke external APIs and custom functions by generating structured JSON payloads based on user intents.

238
Multi-Selectmedium

When implementing grounding in a RAG application using Azure AI Foundry, which TWO practices help minimize hallucinations in model outputs? (Choose TWO)

Select 2 answers
A.Adding explicit system prompt instructions telling the model to refuse answers not supported by context
B.Including source document chunk citations in the prompt context alongside the text
C.Disabling content safety filters to allow unrestricted generation
D.Setting the model temperature parameter to 2.0 to maximize creativity
E.Removing all source context documents from the prompt entirely
AnswersA, B

Negative constraints in system prompts prevent the model from making up facts.

Why this answer

Instructing the model to rely only on context and providing precise retrieved snippets with source citations minimize hallucinations.

239
Multi-Selecteasy

You need to extract key concepts from customer feedback. Which TWO characteristics describe key phrase extraction in Azure AI Language? (Choose two.)

Select 2 answers
A.It evaluates the emotional tone of sentences.
B.It requires a pre-defined taxonomy of thousands of categories.
C.It works well on unstructured text documents.
D.It returns a list of strings denoting the main talking points.
E.It automatically translates documents into English.
AnswersC, D

It is designed to process unstructured text documents effectively.

Why this answer

Key phrase extraction identifies main concepts, works on unstructured text, and returns a list of strings without requiring pre-defined taxonomies.

240
MCQhard

An agent is performing poorly on multi-hop reasoning tasks. What architectural pattern should you implement to improve performance?

A.Increase the temperature to 1.0.
B.Switch to a smaller model to increase speed.
C.Decrease the max_tokens parameter.
D.Use a Chain-of-Thought prompting strategy in the agent's system prompt.
AnswerD

CoT is specifically designed to improve multi-hop reasoning.

Why this answer

Chain-of-thought prompting forces the model to break down complex queries into intermediate logical steps.

241
MCQhard

You are auditing an Azure AI Document Intelligence solution for data privacy compliance. Documents contain sensitive Personally Identifiable Information (PII). When using prebuilt models or custom models, how is data handled by Microsoft regarding model training?

A.Training data uploaded for custom models is made publicly available in the Azure AI gallery after 30 days.
B.Customer data and custom model training data are encrypted at rest and in transit, and are never used to train Microsoft foundation models.
C.Data privacy controls require you to deploy a local container, as cloud APIs store training documents indefinitely.
D.All submitted documents are automatically used to retrain the global prebuilt models overnight.
AnswerB

Azure AI adheres to strict data privacy commitments where customer data is not used for training Microsoft foundational models.

Why this answer

Microsoft does not use customer data submitted to Azure AI Document Intelligence to train baseline prebuilt models or customer custom models.

242
MCQeasy

You need to evaluate the performance of your prompt engineering for a summarization task. Which tool in Azure AI Foundry allows you to run multiple prompt variants against a dataset to compare outputs?

A.Azure OpenAI studio Playground.
B.Azure AI Content Safety.
C.Azure AI Search.
D.Azure AI Prompt Flow.
AnswerD

Prompt flow provides the environment for prompt testing and evaluation.

Why this answer

Prompt flow evaluation allows for comparing prompt versions against specific datasets systematically.

243
MCQeasy

What is the purpose of a 'Resource Group' in the context of Azure AI solutions?

A.To act as a logical container for managing related resources.
B.To provide high availability for the service.
C.To enforce network firewall rules.
D.To store the model weights.
AnswerA

Resource groups group resources for easier management and deployment.

Why this answer

Resource groups act as a logical container for related Azure resources, making it easier to manage the lifecycle, permissions, and billing of a solution.

244
MCQeasy

You are creating a Language resource in the Azure portal. Which blade should you navigate to if you need to retrieve your endpoint URL and subscription keys for your application configuration?

A.Diagnostic settings
B.Pricing tier
C.Keys and Endpoint
D.Access control (IAM)
AnswerC

Keys and Endpoint provides the API keys and resource endpoint URL.

Why this answer

The 'Keys and Endpoint' blade in the Azure portal displays the API keys and endpoint URL required to connect your application to the Azure AI Language service.

245
MCQeasy

You are provisioning an Azure AI Language resource and want to ensure that all telemetry and activity logs are retained for compliance auditing over a 365-day period. Where should you configure this retention policy?

A.In the resource's Diagnostic settings by specifying a retention period on a storage account destination.
B.In the Microsoft Defender for Cloud security policy blade.
C.In the Azure AI Language service Studio configuration panel.
D.In the Azure Cost Management budget alert settings.
AnswerA

Diagnostic settings allow you to define log retention periods in days for storage account destinations.

Why this answer

Diagnostic settings in the Azure portal allow configuring log destinations and retention policies (in days) when sending logs to a Storage Account.

246
MCQeasy

You are configuring an Azure OpenAI Service deployment in Microsoft AI Foundry portal. You need to ensure that the deployed model meets your enterprise compliance requirements for data privacy, ensuring that prompt and completion data are not used to train Microsoft base models. Which deployment type should you select?

A.Global Standard deployment
B.Classic Azure OpenAI fine-tuned model deployment
C.Data Zone Standard deployment with model tuning enabled
D.Provisioned Throughput deployment
AnswerA

Standard deployments guarantee that prompt and completion data are not used for training base models.

Why this answer

Standard deployments in Azure OpenAI Service ensure that customer data is not used to train Microsoft base models. Data-zone residency applies by default to standard deployments in supported regions.

247
Multi-Selecthard

You are evaluating the performance of a custom text classification model in Language Studio. Which THREE evaluation metrics are provided to assess model quality? (Choose three.)

Select 3 answers
A.HTTP request latency in milliseconds
B.F1-score
C.Precision
D.CPU utilization percentage
E.Recall
AnswersB, C, E

F1-score is the harmonic mean of precision and recall.

Why this answer

Custom text classification models are evaluated using Precision, Recall, and F1-score metrics.

248
MCQmedium

You are building an agent using Azure AI Agent Service that needs to maintain long-term conversational memory across multiple sessions with a user. Which feature should you utilize to store and retrieve past user interactions?

A.Agent threads and message history APIs
B.Azure Cache for Redis with session state middleware
C.Azure Cosmos DB vector search collections
D.Azure AI Search index vector store
AnswerA

Threads store conversation state and message history on the service side, allowing agents to maintain context across sessions.

Why this answer

Azure AI Agent Service provides built-in thread management where conversation histories are stored and persisted across sessions.

249
MCQmedium

When using Prompt Flow, you notice that your flow is failing at the 'aggregate' node. What is the most likely cause?

A.The input data from the previous node does not match the expected schema of the aggregate node.
B.The system prompt is too long.
C.The search index is empty.
D.The model is too busy.
AnswerA

Prompt flow nodes require strict input/output alignment.

Why this answer

The aggregate node expects a specific schema and iteration results from the previous node; if these are missing or mismatched, the flow fails.

250
MCQhard

You are designing a RAG solution in Azure AI Foundry where sensitive enterprise documents must be retrieved. Users should only see search results corresponding to their security clearance level. How should you implement document-level access control in Azure AI Search?

A.Store user security groups in document metadata and apply OData filter expressions during search queries
B.Configure Microsoft Entra ID authentication on the Azure OpenAI endpoint
C.Encrypt the search index using customer-managed keys (CMK) in Azure Key Vault
D.Use Azure AI Content Safety to redact sensitive keywords before indexing
AnswerA

OData filters dynamically restrict search results to documents matching the user's security clearance metadata.

Why this answer

Security trimming in Azure AI Search is implemented by adding user security identifiers (IDs or group memberships) as metadata fields on each document and applying OData filter expressions at query time.

251
MCQmedium

Your enterprise customer service department receives thousands of support emails daily. Management wants a system that ingests long support tickets and automatically outputs a concise summary highlighting the most critical issues. Which Azure AI Language service feature should you implement?

A.Named Entity Recognition (NER)
B.Text summarization (Abstractive or Extractive)
C.Key phrase extraction
D.Opinion mining
AnswerB

Text summarization analyzes the content and condenses long documents into summaries using either extractive or abstractive methods.

Why this answer

Extractive or abstractive text summarization is designed to process long documents and generate concise summaries.

252
MCQmedium

You are building an agent in Azure AI Foundry using a prompt flow. You need to ensure that the agent only uses company-approved documentation for its responses. Which feature should you implement to meet this requirement?

A.Enable prompt tuning
B.Implement system message constraints
C.Configure Grounding with data
D.Deploy a fine-tuned model
AnswerC

Grounding with data allows you to connect your agent to specific data sources like Azure AI Search to restrict the scope of answers.

Why this answer

Grounding with data is the primary mechanism in Azure AI Foundry to limit an agent's knowledge to specific data sources.

253
MCQhard

You are deploying a Custom Vision object detection model as a Docker container on an Azure Kubernetes Service (AKS) cluster. During load testing, you notice high memory consumption and container restarts. Which configuration setting in the deployment manifest should you review to ensure stability?

A.Billing endpoint URL syntax.
B.Azure Blob Storage connection string persistence.
C.Custom Vision iteration publishing name environment variable.
D.Kubernetes resource requests and limits for memory and CPU.
AnswerD

Configuring adequate memory and CPU limits prevents OOM termination and CPU throttling for vision containers.

Why this answer

Custom Vision container deployments require proper CPU and memory limits/requests configured in the Kubernetes pod spec to prevent Out-Of-Memory (OOM) kills during heavy inference workloads.

254
MCQmedium

You are deploying an open-source model (such as Llama 3) via Model Catalog in Azure AI Foundry. You want to optimize inference latency and throughput for high-concurrency requests. Which deployment type should you select?

A.Azure SQL Database scalar User Defined Functions
B.Local developer workstation Docker container
C.Azure Functions consumption plan
D.Managed compute with serverless API deployment
AnswerD

Serverless API deployments in Azure AI Foundry offer managed infrastructure optimized for open-source model inference throughput.

Why this answer

Serverless APIs with pay-as-you-go billing or managed compute provisioned through Azure AI Foundry Model Catalog provide optimized inference hosting.

255
MCQhard

You are using the Azure AI Language REST API to perform extractive summarization on a large collection of research papers. You want to prioritize the top sentences that represent the core findings. Which parameter can you adjust in the API request to control the maximum number of sentences returned in the summary?

A.summaryLengthLimit
B.topKeyPhrases
C.maxSentenceCount
D.minConfidenceScore
AnswerC

maxSentenceCount dictates the maximum number of extracted sentences in extractive summarization.

Why this answer

In extractive summarization requests, you can specify the 'maxSentenceCount' parameter to limit how many sentences appear in the summary output.

256
Multi-Selectmedium

Which TWO image file formats are natively supported by Azure AI Vision for image analysis? (Choose two.)

Select 2 answers
A.MP3
B.MKV
C.WAV
D.JPEG
E.PNG
AnswersD, E

JPEG is fully supported by Azure AI Vision.

Why this answer

Azure AI Vision supports standard web formats including JPEG, PNG, GIF, BMP, WEBP, and TIFF (for Read).

257
Multi-Selecthard

You are configuring tool integration for an agent. Which THREE aspects must be clearly defined in the tool definition for the agent to correctly use it?

Select 3 answers
A.The tool's description, explaining what it does and when to call it.
B.The total number of requests allowed per hour.
C.The authentication credentials (hardcoded).
D.The tool's name.
E.The JSON schema for the function arguments.
AnswersA, D, E

Descriptions enable the model's 'reasoning' to select the right tool.

Why this answer

Tools require a name, a clear description for the model, and a valid parameter schema to function correctly.

258
MCQhard

You are designing a multimodal solution that combines Azure AI Vision dense captions with Azure OpenAI GPT-4 Vision. You want to extract precise bounding box coordinates for multiple objects in complex scenes and pass them as structured context. Which API feature should you invoke?

A.features=denseCaptions
B.features=objects
C.features=smartCrops
D.features=tags
AnswerA

Dense captions provide detailed descriptions and bounding box coordinates for multiple sub-regions in an image.

Why this answer

Dense captions return a set of regional bounding boxes and corresponding descriptive text snippets.

259
MCQeasy

You are using Azure AI Vision to analyze an image. You want to receive the output in Spanish instead of English. Which query parameter should you add to your request?

A.language=es
B.translation=spanish
C.output-lang=es
D.locale=es-ES
AnswerA

The language parameter translates or returns analysis results in the specified ISO language code.

Why this answer

The 'language' query parameter specifies the language for the output results (e.g., language=es for Spanish).

260
MCQmedium

When labeling documents for a custom extraction model, what happens if you label the same field inconsistently across documents?

A.Model accuracy will decrease
B.The training will fail immediately
C.No impact
D.The model will automatically correct the labels
AnswerA

Inconsistency directly impacts model learning and performance.

Why this answer

Inconsistent labeling introduces noise into the training data, which reduces the final F1 score of the model.

261
Multi-Selectmedium

Which THREE features are provided by the prebuilt Layout model in Azure AI Document Intelligence? (Choose THREE)

Select 3 answers
A.Automatic sentiment analysis of paragraph text
B.Extraction of text lines and words with bounding polygons
C.Table structure extraction (rows, columns, cell spans)
D.Language translation of extracted text into English
E.Selection mark state detection (selected or unselected checkboxes)
AnswersB, C, E

The layout model extracts detailed text elements and their bounding regions.

Why this answer

The layout model extracts text lines and words, tables, and selection marks.

262
MCQeasy

You are designing a solution that processes user reviews. You want to extract key terms such as 'battery life', 'screen quality', and 'customer service' from the text. Which Azure AI Language feature is best suited for this?

A.Key phrase extraction
B.Sentiment analysis
C.PII detection
D.Language detection
AnswerA

Key phrase extraction identifies key terms and concepts in unstructured text.

Why this answer

Key phrase extraction extracts the most important terms and phrases from input documents.

263
MCQmedium

You are integrating Azure OpenAI into an enterprise application. You need to prevent prompt injection attacks where users attempt to override the system instructions. Which approach provides the most robust defense?

A.Implement an input guardrail using Azure AI Content Safety to detect and block jailbreak attempts before they reach the model.
B.Set the model frequency penalty to 2.0 to penalize repeated malicious phrases.
C.Append 'Ignore previous instructions' at the end of every user message.
D.Encode all user input in Base64 before sending it to the model.
AnswerA

Azure AI Content Safety provides dedicated jailbreak and prompt injection detection classifiers.

Why this answer

Using dual-prompt architectures or Azure AI Content Safety groundedness checks provides robust detection and mitigation against prompt injections.

264
MCQhard

You are configuring a secure enterprise architecture where Azure AI Vision calls must not traverse the public internet. You create a private endpoint for your Azure AI Services resource in your Virtual Network. Which DNS configuration step is required for client applications within the VNet to resolve the private endpoint correctly?

A.Add the endpoint IP to the local hosts file of every Azure AI Vision instance.
B.Configure Azure Private DNS zones corresponding to the cognitive services DNS zone and link it to your VNet.
C.Configure public DNS records on your corporate DNS server to point to the public IP.
D.Modify the routing table to drop all DNS traffic.
AnswerB

Private DNS zones ensure that DNS queries within the VNet resolve the service endpoint to its assigned private IP address.

Why this answer

When configuring private endpoints for Azure AI Services, you must update or create Private DNS zones (e.g., `privatelink.cognitiveservices.azure.com`) to map the resource FQDN to the private IP address.

265
Multi-Selectmedium

You are troubleshooting a deployment failure where an ARM template fails to provision an Azure OpenAI account. Which TWO potential causes should you investigate? Each correct answer presents part of the solution.

Select 2 answers
A.The subscription has exceeded its regional quota or model deployment quota limit.
B.The local workstation IP address is not whitelisted in Azure DevOps.
C.The client web browser cache is corrupted.
D.The requested region does not support Azure OpenAI model deployments.
E.The local git repository has uncommitted changes.
AnswersA, D

Exceeding subscription quotas for Azure OpenAI results in provisioning errors.

Why this answer

Quota limitations and region availability are common causes of OpenAI deployment failures.

266
Multi-Selecthard

Which THREE attributes are critical for a successful custom model training set?

Select 3 answers
A.High-quality images
B.Consistent labeling across all documents
C.Representative samples of all document layouts
D.All documents from the same printer
E.At least 1,000 pages per model
AnswersA, B, C

Essential for accuracy.

Why this answer

Training sets must be high-quality, representative, and clearly labeled to achieve high model performance.

267
Multi-Selectmedium

When building a custom template model in Azure AI Document Intelligence Studio, which THREE types of data or attributes can you extract by labeling your training documents? (Choose THREE)

Select 3 answers
A.Facial biometric embeddings
B.Real-time audio transcription timestamps
C.Tables (including row and column structures)
D.Selection marks (checkboxes and radio buttons)
E.Fields (key-value pairs)
AnswersC, D, E

Custom tables can be labeled to extract line items and grid data.

Why this answer

Custom template models allow labeling fields, tables, and selection marks.

268
MCQeasy

You are developing a global news aggregation application using the Azure AI Language service. You need to automatically identify the primary language of incoming user-submitted articles across 120 different languages before routing them to translation pipelines. Which API endpoint should you call?

A.POST /language/translate?api-version=3.0
B.POST /language/:analyze-text?api-version=2023-04-01 with the CustomEntityRecognition task
C.POST /language/analyze-text/jobs?api-version=2023-04-01 with the TextSentimentAnalysis task
D.POST /language/:analyze-text?api-version=2023-04-01 with the TextLanguageDetection task
AnswerD

This is the correct synchronous analyze-text API endpoint and task name for language detection.

Why this answer

The Language detection feature of the Azure AI Language service evaluates text input and returns the detected language name, ISO 639-1 code, and a confidence score for up to 120 languages.

269
MCQeasy

You are developing an application to extract specific fields from invoices using Azure AI Document Intelligence. Which prebuilt model should you use for this task?

A.custom-template
B.prebuilt-layout
C.prebuilt-document
D.prebuilt-invoice
AnswerD

This model is specialized for extracting invoice-specific data fields.

Why this answer

The 'prebuilt-invoice' model is specifically designed to extract key-value pairs and line items from invoices.

270
MCQmedium

You are building a chat application with Azure OpenAI. To improve perceived latency for users, you want to stream the model's response token-by-token to the client interface. Which API parameter or client SDK feature should you enable?

A.Set max_tokens to 1
B.Configure Azure AI Search with hybrid search
C.Enable semantic caching
D.Enable streaming in the chat completion request (stream=True)
AnswerD

Streaming returns chunks of text as Server-Sent Events (SSE), drastically improving user experience and perceived response speed.

Why this answer

Enabling streaming mode (`stream=True`) in the Azure OpenAI chat completion API allows tokens to be returned incrementally as they are generated.

271
MCQmedium

Your application uploads an image to Azure AI Vision for analysis. You need to generate thumbnail images with specific aspect ratios while maintaining the primary region of interest. Which API endpoint or feature should you use?

A.Use Custom Vision Training API to train a cropping model.
B.Call the Analyze Image API with features=objects and crop the image locally.
C.Call the Get Thumbnail API passing width, height, and smartCropping parameters.
D.Use Azure AI Document Intelligence layout analysis.
AnswerC

Get Thumbnail analyzes the image and crops it around the region of interest.

Why this answer

The 'smartCrops' feature in Image Analysis 4.0 generates thumbnail coordinates optimized for target aspect ratios.

272
MCQhard

You are designing a monitoring solution for an Azure AI Document Intelligence resource that processes thousands of invoices daily. You need to track custom metric alerts when HTTP 429 (Too Many Requests) errors exceed a specific threshold over a 5-minute window. Where should you configure this alert?

A.Application Insights availability web tests
B.Microsoft Defender for Cloud threat protection policies
C.Azure Monitor metric alerts
D.Azure AI Studio diagnostics dashboard
AnswerC

Azure Monitor metric alerts natively track the 'Total Calls' and error status codes like HTTP 429 emitted by Azure AI service resources.

Why this answer

Azure Monitor metric alerts allow you to evaluate platform metrics such as 'Total Calls' or 'HTTP 429' against dynamic or static thresholds.

273
MCQmedium

You are integrating Azure AI Document Intelligence into a web application. You call the analyzeDocument API and receive a 202 Accepted response. What is the correct pattern for your application to retrieve the final extraction results?

A.Wait for exactly 60 seconds and then send a GET request to the original analyze endpoint.
B.Immediately parse the response body of the 202 Accepted response for the extracted JSON data.
C.Poll the URL provided in the Operation-Location response header until the status returns 'succeeded', then fetch the results.
D.Open a WebSocket connection to the streaming endpoint specified in the WebSocket-Location header.
AnswerC

Asynchronous operations require polling the Operation-Location endpoint until completion.

Why this answer

Document Intelligence analysis is an asynchronous operation. You must poll the Operation-Location URL returned in the response header until the status is succeeded.

274
Multi-Selectmedium

Which TWO actions should you take to ensure your agentic solution is 'Responsible'?

Select 2 answers
A.Enable Content Safety filters for all inputs and outputs.
B.Disable all logging to protect user privacy.
C.Provide transparent disclosures to users that they are interacting with an AI agent.
D.Set the model temperature to 1.0 to increase diversity.
E.Allow the agent to access the public internet without restrictions.
AnswersA, C

This is a mandatory technical guardrail.

Why this answer

Responsible AI involves both technical guardrails (Content Safety) and transparency (disclosures/citations).

275
MCQmedium

You are deploying an LLM in Azure AI Foundry and want to restrict access to the endpoint to a specific virtual network. What should you configure?

A.Azure AD Conditional Access
B.Firewall Rule on the Model
C.Private Endpoint
D.Managed Identity
AnswerC

Private endpoints provide secure network access.

Why this answer

Private Endpoints are used to secure access to Azure services within a VNet.

276
Multi-Selecthard

You are conducting a security review of an Azure AI Search service and Azure OpenAI integration. You discover potential security misconfigurations. Which THREE remediation actions should you take to align with Azure security benchmarks? Each correct answer presents part of the solution.

Select 3 answers
A.Grant Contributor permissions to anonymous external users.
B.Enable diagnostic logging and configure alerts for unauthorized access attempts and HTTP 401/403 errors.
C.Disable public network access on services where private endpoints are fully implemented.
D.Expose admin API keys in client-side single-page applications for direct access.
E.Rotate all static API keys regularly and transition client applications to Microsoft Entra ID managed identities.
AnswersB, C, E

Monitoring authentication failures helps detect security breaches or misconfigured clients.

Why this answer

Disabling public access, rotating keys, and using managed identities align with security benchmarks.

277
MCQhard

You are configuring an Azure AI Document Intelligence custom extraction pipeline that processes thousands of diverse medical forms daily. You notice that extraction confidence scores for certain hand-written fields are low. What is the most effective remediation step within the Document Intelligence workflow?

A.Disable the OCR engine parameter in the API call to bypass text recognition steps.
B.Decrease the model training iterations configuration setting to prevent overfitting on handwriting.
C.Add additional labeled training documents that specifically feature diverse handwriting samples for those low-confidence fields.
D.Switch from a Custom Neural model to a Custom Template model.
AnswerC

Expanding the training dataset with representative variations helps the neural network learn diverse writing patterns, increasing accuracy and confidence.

Why this answer

Adding more high-quality, representative labeled training samples containing diverse handwriting styles directly improves custom model performance and confidence on ambiguous fields.

278
MCQhard

You are troubleshooting a custom text classification model that exhibits high training loss and low validation accuracy, indicating severe model underfitting. What adjustment should you make to improve the model?

A.Lower the confidence score threshold to 0.01.
B.Increase the diversity and volume of training data and ensure model training is allowed to complete fully.
C.Reduce the training dataset size by 90%.
D.Switch the resource pricing tier from Standard to Free.
AnswerB

Underfitting is addressed by providing more training data and ensuring adequate model learning capacity.

Why this answer

Underfitting occurs when a model is too simple or hasn't learned enough patterns. Increasing training iterations (epochs) or adding more descriptive features and training data helps resolve underfitting.

279
MCQeasy

Which of the following is a key advantage of using a 'prebuilt' model over a 'custom' model?

A.Higher accuracy for niche documents
B.Lower cost per page
C.Ability to define custom fields
D.No training required
AnswerD

Prebuilt models are pre-trained by Microsoft.

Why this answer

Prebuilt models are ready to use immediately without the need for training data.

280
Multi-Selectmedium

You are optimizing an Azure OpenAI deployment in Azure AI Foundry for high-throughput batch processing. Which TWO strategies or configurations should you consider? (Choose TWO)

Select 2 answers
A.Azure OpenAI Batch API for asynchronous high-volume jobs at reduced costs
B.Disabling SSL certificate validation in the client SDK
C.Provisioned Throughput Units (PTUs) for guaranteed model capacity and predictable latency
D.Setting model max_tokens to 1
E.Routing all traffic through public developer local workstations
AnswersA, C

The Batch API processes large volumes of asynchronous requests with dedicated quota and cost benefits.

Why this answer

Batch API usage and provisioned throughput units (PTUs) are designed for managing high-volume enterprise throughput and batch workloads.

281
Multi-Selectmedium

Which TWO actions should you perform to troubleshoot a model that consistently misinterprets tables?

Select 2 answers
A.Use the Read API instead
B.Disable the layout model
C.Change the resource region
D.Verify labels for alignment and completeness
E.Retrain with more diverse training samples
AnswersD, E

Incorrect labeling is the most common cause of errors.

Why this answer

Labeling variations of the table and ensuring high image quality are the primary troubleshooting steps.

282
MCQmedium

You are developing an agent using Azure AI Agent Service and notice that the model consistently refuses to answer queries about company-internal technical specifications despite the documentation being uploaded to the vector index. Which action should you prioritize?

A.Enable 'Content Safety' to bypass filtering.
B.Increase the system prompt length in the model configuration.
C.Verify the 'Grounding' settings in the Azure AI Foundry portal to ensure the search index is correctly linked.
D.Switch the model to a higher parameter count version.
AnswerC

Grounding settings connect the agent to the specific data index required for RAG.

Why this answer

Grounding ensures the model queries the provided data before generating a response. If it refuses to answer, the grounding configuration is likely failing or missing the retrieval step.

283
MCQmedium

Your organization is running multiple Azure AI services. You need to configure alerts so that when the availability of any AI service drops below 99.% over a 15-minute window, an incident is created in ServiceNow. Which Azure Monitor feature should you use?

A.Azure Cost Management alerts
B.Azure Monitor metric alerts combined with Action Groups configured with a Webhook receiver
C.Azure Advisor performance recommendations
D.Microsoft Defender for Cloud threat detection
AnswerB

Metric alerts monitor availability metrics, and Action Groups can forward alerts to external ITSM systems like ServiceNow via webhooks.

Why this answer

Azure Monitor metric alerts integrated with Action Groups allow sending notifications to webhooks (such as ServiceNow).

284
MCQeasy

Which prebuilt Azure AI Document Intelligence model should you use to extract line items, merchant name, transaction date, and total tax from point-of-sale purchase receipts?

A.prebuilt-invoice
B.prebuilt-idDocument
C.prebuilt-businessCard
D.prebuilt-receipt
AnswerD

The receipt model extracts transaction details, merchant info, and line items from receipts.

Why this answer

The prebuilt receipt model (prebuilt-receipt) is specifically trained to extract data from sales receipts.

285
Multi-Selectmedium

Which THREE metrics should you monitor to evaluate the effectiveness of a RAG-based agent?

Select 3 answers
A.Retrieval Precision (the accuracy of the retrieved documents).
B.Answer Relevance (how well the answer matches the user query).
C.Model Parameter count.
D.The number of times the model was restarted.
E.Model Latency (time to first token).
AnswersA, B, E

This measures how good the search results are.

Why this answer

RAG evaluation requires metrics for retrieval precision, answer relevance, and factual grounding.

286
MCQmedium

You are designing a generative AI application that processes user uploaded documents. You must ensure that Personally Identifiable Information (PII) such as Social Security numbers are automatically redacted before the text is sent to the Azure OpenAI model. Which service should you integrate into your pipeline?

A.Azure AI Language service (PII detection and redaction)
B.Azure AI Translator
C.Azure AI Vision OCR
D.Azure Monitor Application Insights
AnswerA

Azure AI Language can identify and redact sensitive personal data prior to model inference.

Why this answer

Azure AI Language provides dedicated features for PII detection and redaction (named entity recognition for PII).

287
MCQmedium

You are configuring text summarization using the Azure AI Language REST API. You need to generate a summary that consists of a few distinct sentences pulled directly from the source document without altering their wording. Which summarization kind should you specify?

A.Key phrase extraction
B.Extractive summarization
C.Abstractive summarization
D.Custom single-label classification
AnswerB

Extractive summarization produces a summary by taking verbatim sentences from the source text.

Why this answer

Extractive summarization extracts sentences that collectively capture the most important context from the source document without rewriting them.

288
Multi-Selectmedium

You are preparing training data for a custom named entity recognition (NER) project in Language Studio. Which THREE guidelines should you follow to ensure high model accuracy? (Choose three.)

Select 3 answers
A.Ensure consistent and accurate tagging of entities across all training documents.
B.Use exactly one training document for the entire project.
C.Provide sufficient examples for each entity type based on service guidelines.
D.Include a diverse set of training documents that reflect real-world data distribution.
E.Tag every single word in every document as an entity.
AnswersA, C, D

Consistency in labeling is critical for model training accuracy.

Why this answer

Best practices for custom NER training data include providing a diverse set of documents, ensuring consistent tagging across all instances, and maintaining sufficient sample counts per entity type.

289
MCQmedium

Your development team wants to deploy a new version of an Azure OpenAI model while ensuring zero downtime for client applications currently calling the existing model deployment. What is the recommended deployment strategy?

A.Overwrite the existing model deployment name with the new model version.
B.Delete the Azure OpenAI resource and provision a new one in the same region.
C.Create a new deployment with a distinct deployment name for the new model version, then update application configuration to point to the new deployment.
D.Downgrade the pricing tier to pause traffic while updating the model version.
AnswerC

Using a distinct deployment name allows side-by-side deployment and seamless cutover.

Why this answer

Blue-green or side-by-side deployments allow deploying the new model version alongside the old one, updating client application endpoints or routing rules after validation.

290
MCQhard

You are implementing an agentic workflow where an agent executes tools that modify corporate data. To adhere to Responsible AI principles, you must ensure a human reviewer approves sensitive tool actions before they are executed. What design pattern should you implement in Azure AI Agent Service?

A.Fully autonomous agent loop with max iterations set to 50
B.Human-in-the-loop (HITL) approval workflow for tool execution
C.Azure API Management rate limiting policy
D.Asynchronous event grid notification after tool execution
AnswerB

HITL pauses the agent's tool execution loop, presenting the proposed action to a human reviewer for approval.

Why this answer

Human-in-the-loop (HITL) review pauses agent execution before tool calls that perform write or destructive actions, awaiting explicit human approval.

291
Multi-Selectmedium

Which TWO actions should you take to ensure that your Azure AI deployment is secure?

Select 2 answers
A.Keep the public IP exposed at all times.
B.Use Managed Identities for service-to-service communication.
C.Store all keys in a public GitHub repository.
D.Always use the latest TLS version for all communication.
E.Use the default 'Owner' role for all users.
AnswersB, D

Managed identities eliminate the need for hardcoded credentials.

Why this answer

Enforcing TLS 1.2+ and using managed identities are standard security best practices for Azure AI.

292
MCQeasy

Which of the following is a core principle of Responsible AI in Azure AI Foundry?

A.Bypassing all safety filters.
B.Maximizing cost savings at any cost.
C.Forcing the model to agree with the user.
D.Transparency and explainability.
AnswerD

Users should know when they are interacting with AI and understand the basis of its responses.

Why this answer

Transparency is a core principle, often implemented by providing sources and citations in AI responses.

293
MCQhard

You are analyzing legal contracts using Azure AI Language custom named entity recognition. You notice that overlapping entities (e.g., 'Acme Corporation' inside a larger entity 'Acme Corporation Headquarters') are causing extraction errors in your custom model. How should you resolve this in Language Studio?

A.Configure the project schema to support overlapping entities during project creation.
B.Increase the confidence score threshold to 0.99.
C.Switch the project type from custom NER to custom text classification.
D.Convert all contract texts to lowercase before tagging.
AnswerA

Custom NER projects require enabling overlapping entities settings at project creation if nested or intersecting tags occur.

Why this answer

When configuring custom NER projects, you can enable or configure support for overlapping entities during project creation if your schema requires nested or overlapping extractions.

294
MCQeasy

In Azure AI Foundry, which tool would you use to measure the quality of your RAG application against a golden dataset?

A.Evaluation
B.Model Catalog
C.Deployment
D.Prompt Flow Tracing
AnswerA

Evaluation measures performance metrics.

Why this answer

Evaluation in Azure AI Foundry uses metrics to assess RAG pipelines.

295
MCQeasy

You want to extract gender and age attributes from human faces in photos. While Azure AI Face service handles facial analysis, which base Azure AI Vision 4.0 feature detects the presence and bounding boxes of people?

A.features=objects
B.features=people
C.features=tags
D.features=read
AnswerB

The people feature detects human bounding boxes in Image Analysis 4.0.

Why this answer

The 'people' feature in Image Analysis 4.0 detects humans and returns their bounding boxes.

296
MCQmedium

Your company's security policy requires that all Azure AI Services accounts disable public network access completely. However, internal applications hosted inside a specific Azure Virtual Network must still communicate with the AI services. How should you configure network access?

A.Configure an inbound Network Security Group (NSG) rule on the AI resource subnet.
B.Disable public network access and configure a Private Endpoint for the resource.
C.Keep public network access enabled and rely on API keys for security.
D.Deploy a VPN gateway inside the Azure AI resource group.
AnswerB

Setting public network access to 'Disabled' blocks all public internet access, while Private Endpoints maintain secure VNet connectivity.

Why this answer

Disabling public network access while enabling private endpoints ensures that traffic flows exclusively through the virtual network's private IP space.

297
MCQeasy

You are provisioning a new Azure AI Vision resource in the Azure Portal. You need to choose between a 'F0' free pricing tier and an 'S0' standard pricing tier. What is a key limitation of selecting the F0 pricing tier?

A.It does not support regional data residency guarantees.
B.It disables diagnostic log streaming to Azure Monitor.
C.It restricts the total number of calls allowed per month and limits transaction rates.
D.It prohibits the use of managed identities for authentication.
AnswerC

F0 is restricted to a specific low monthly call quota and limited transaction-per-second limits.

Why this answer

The F0 pricing tier restricts the total number of transactions allowed per month and limits concurrency.

298
MCQeasy

You are developing a .NET application that analyzes customer feedback. Which NuGet package should you install to use the Azure AI Language client library in C#?

A.Azure.AI.Language.Text
B.Microsoft.CognitiveServices.TextAnalytics
C.Azure.AI.TextAnalytics
D.Microsoft.Azure.CognitiveServices.Language
AnswerC

Azure.AI.TextAnalytics is the current client library NuGet package.

Why this answer

The official NuGet package for Azure AI Language text analysis in .NET is `Azure.AI.TextAnalytics`.

299
MCQhard

You are building an advanced agentic RAG solution. When a user asks a complex question, the agent needs to generate multiple search queries, execute them in parallel against Azure AI Search, synthesize the findings, and then formulate a final answer. Which agent pattern are you implementing?

A.Model fine-tuning
B.Single-shot keyword extraction
C.Static prompt templating
D.Query decomposition and parallel retrieval
AnswerD

Decomposition breaks complex prompts into targeted sub-queries, improving recall across diverse document sources.

Why this answer

Query decomposition and parallel retrieval is an advanced RAG pattern where complex queries are broken into sub-queries and executed concurrently.

300
MCQmedium

You are developing an Azure AI Search index for a RAG application. You need to ensure that the search results include semantic relevance scores to improve the quality of the model's grounding. Which feature must you enable?

A.Use Custom Analyzers
B.Enable Knowledge Store
C.Enable Semantic Ranker on the search service
D.Enable Hybrid Search with vectorization
AnswerC

Semantic ranker is required for semantic reranking.

Why this answer

Semantic ranker is required to provide semantic relevance scores (semantic search) to rerank results.

Page 3

Page 4 of 7

Page 5

All pages