Courseiva

Microsoft Certified: Azure AI Apps and Agents Developer Associate (AI-103) (AI-103) (AI-103) — Questions 151225

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

Page 2

Page 3 of 7

Page 4
151
MCQhard

You are designing a multi-agent solution in Azure AI Foundry where agents collaborate asynchronously. One agent produces an intermediate JSON payload that must be strictly validated before being consumed by the next agent. What feature should you implement?

A.Model temperature set to 1.0
B.Azure API Management XML transformation policy
C.Structured Outputs with strict JSON Schema enforcement
D.Azure AI Content Safety hate speech detection
AnswerC

Structured Outputs guarantee that the model response conforms 100% to the provided JSON Schema.

Why this answer

Structured Outputs (JSON mode with strict schema enforcement) ensures that model completions conform precisely to a defined JSON Schema.

152
MCQhard

You are training a Custom Vision object detection model. You have 1,000 images, but only 200 of them contain the rare object you want to detect. The other 800 images are background images without the object. What impact will this class imbalance have on training, and how should you address it?

A.Reduce training iteration time to 1 minute.
B.Convert the project to multilabel classification.
C.Ensure that you have tagged at least 15 bounding boxes per tag; having background images is beneficial for reducing false positives.
D.Delete all 800 background images immediately, as they will cause training to fail.
AnswerC

Background images act as negative examples to reduce false positives, provided the positive tag has met the minimum instance threshold.

Why this answer

While having background (negative) images helps reduce false positives, an extreme imbalance or lack of positive bounding box instances per image can hinder learning. Ensuring sufficient positive instances (at least 15 per tag, preferably more for robust object detection) is essential.

153
MCQmedium

You are processing medical intake forms using Azure AI Language. You need to identify and categorize medical terms, medications, and dosages. Which feature should you use?

A.Health text analytics
B.Key phrase extraction
C.Standard Named Entity Recognition (NER)
D.Custom text classification
AnswerA

Health text analytics is designed to extract and label medical-specific entities and assertions.

Why this answer

The health text analytics feature (part of Azure AI Language) extracts medical entities and relations specifically tailored for healthcare domains.

154
Multi-Selectmedium

Which TWO of the following are prerequisites for training a custom extraction model?

Select 2 answers
A.A set of labeled data in JSON format
B.An Azure SQL database
C.A pre-existing template model
D.A pre-installed local labeling agent
E.An Azure Blob Storage container with documents
AnswersA, E

Labels are essential for training custom models.

Why this answer

You need a source of files in Azure Blob Storage and the files must be associated with valid labels (via .labels.json).

155
MCQeasy

Your team needs to extract the most important talking points from research papers using the Azure AI Language service. You want the service to return a list of salient terms and phrases from the input document. Which API task should you execute?

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

Key phrase extraction evaluates the input text and returns a list of important semantic concepts and phrases.

Why this answer

Key phrase extraction quickly identifies the main concepts in a body of text.

156
MCQeasy

You are building an image search engine. You want to extract visual feature tags from user photos using Azure AI Vision 4.0. Which visual feature should you request?

A.features=read
B.features=caption
C.features=smartCrops
D.features=tags
AnswerD

The tags feature extracts keywords describing the visual content of the image.

Why this answer

The 'tags' visual feature extracts content-based tags or keywords representing objects, scenery, and concepts in the image.

157
MCQmedium

You are developing a RAG solution using Azure AI Search. Users are searching using domain-specific acronyms that are not well-represented in standard embedding models, resulting in poor retrieval scores. What feature should you enable in Azure AI Search to improve keyword matching for exact acronyms?

A.Increasing the embedding dimension size to 3072
B.Enabling Azure AI Content Safety filters
C.Setting the semantic ranker threshold to zero
D.Hybrid search combining vector search and full-text BM25 search
AnswerD

Hybrid search merges vector embeddings with keyword BM25 scoring, capturing both semantic intent and exact keyword matches like acronyms.

Why this answer

Hybrid search combines vector search (for semantic meaning) with full-text keyword search (BM25 algorithm), ensuring exact acronyms and part numbers are matched reliably.

158
MCQmedium

You are building an agent in Azure AI Foundry that needs to retrieve information from a company SharePoint site. Which component should you implement to ground the agent?

A.Blob Storage integration
B.Azure AI Search with a SharePoint connector
C.Direct REST API call to SharePoint
D.Microsoft Graph Data Connect directly to the model
AnswerB

The SharePoint connector enables indexing for retrieval.

Why this answer

Azure AI Search with a SharePoint indexer is the standard way to ground agents in external document stores.

159
Multi-Selectmedium

You are evaluating generative AI model outputs in Azure AI Foundry. Which TWO automated evaluation metrics assess the quality and correctness of RAG responses against source data? (Choose TWO)

Select 2 answers
A.Groundedness (measuring adherence to source context)
B.Virtual network gateway throughput
C.Storage account transaction count per hour
D.CPU utilization percentage of the Azure App Service plan
E.Relevance (measuring how well the response addresses the user query)
AnswersA, E

Groundedness evaluates whether model answers are supported by retrieved context.

Why this answer

Groundedness and relevance are core Azure AI Foundry automated evaluation metrics used to assess RAG application quality.

160
MCQeasy

You are planning the disaster recovery strategy for an Azure AI Translator resource. In the event of a regional outage in East US, you want to ensure minimal downtime and continuity of service. Which deployment model should you select?

A.Deploy a primary resource in East US and a secondary resource in the paired region (West US), then implement client-side failover logic.
B.Configure an Azure Backup vault to snapshot the AI model weights daily.
C.Store all translation requests in Azure Blob Storage and replay them when the region recovers.
D.Rely on Azure's native synchronous cross-region database replication for cognitive services.
AnswerA

Active-passive or active-active multi-region deployments with client-side retries or Azure Traffic Manager provide reliable business continuity.

Why this answer

Deploying paired or secondary multi-service resources in a secondary paired region allows applications to failover to the secondary endpoint.

161
MCQeasy

You are designing a generative AI application in Azure AI Foundry and want to categorize user queries into predefined intent classes before routing them to specific agents. Which prompt engineering technique should you apply?

A.Zero-shot random generation
B.Enabling JSON output mode without schema definition
C.Few-shot prompting with labeled classification examples
D.Increasing model temperature to 1.5
AnswerC

Providing input-output classification examples in the prompt teaches the model how to categorize incoming queries.

Why this answer

Few-shot classification prompting provides examples of inputs paired with their corresponding categories to guide the model's classification accuracy.

162
Multi-Selecthard

Which THREE features are critical for managing the lifecycle of an AI model in a production environment?

Select 3 answers
A.Model registry for version control.
B.Hardcoding all environment settings.
C.CI/CD integration for automated deployments.
D.Manual file copying via FTP.
E.Real-time monitoring of inference metrics.
AnswersA, C, E

Versioning is essential for tracking model iterations.

Why this answer

Versioning, monitoring, and automated deployment pipelines are essential for production AI lifecycle management.

163
Multi-Selecthard

You are optimizing the cost of your Azure AI infrastructure. You notice several resources are underutilized or incorrectly provisioned. Which THREE actions should you take to reduce unnecessary spending? Each correct answer presents part of the solution.

Select 3 answers
A.Downgrade development and test resources from high-tier production SKUs to free or lower standard tiers where appropriate.
B.Provision dedicated Azure OpenAI PTUs for idle development sandboxes.
C.Identify and delete abandoned or unused Azure AI resource accounts.
D.Implement Azure Cost Management budgets and anomaly detection alerts.
E.Scale out search service partitions and replicas to maximum capacity permanently.
AnswersA, C, D

Using lower SKUs for non-production environments lowers monthly operational costs.

Why this answer

Deleting idle resources, right-sizing tiers, and setting budgets reduce costs effectively.

164
MCQeasy

When engineering prompts for Azure OpenAI models, you want to guide the model's output format and persona effectively. Which practice represents the most reliable prompt engineering technique?

A.Embedding all context and instructions exclusively inside user messages without utilizing system messages.
B.Providing clear role definition, explicit formatting instructions, and few-shot examples within the system message.
C.Using overly complex, ambiguous phrasing to encourage the model to explore diverse latent spaces.
D.Relying entirely on zero-shot inference without system messages to maximize model creativity.
AnswerB

Structuring prompts with clear system instructions and few-shot examples significantly improves adherence to desired formats and behaviors.

Why this answer

Providing clear instructions, role definition, and formatting constraints along with few-shot examples yields the most consistent and accurate model behavior.

165
MCQhard

You are managing an enterprise Azure AI Search deployment. During an audit, you notice that index data is encrypted at rest using Microsoft-managed keys. Management mandates that you switch to customer-managed keys (CMK) stored in Azure Key Vault immediately without losing index data or rebuilding the service. How should you implement this?

A.Delete the search service and provision a brand new one with CMK enabled from inception.
B.Modify the DNS records to point to a new CMK-encrypted endpoint.
C.Configure encryption with customer-managed keys in the Azure portal on the existing search service instance and assign necessary Key Vault access permissions.
D.Run an Azure Automation runbook that exports indices to blob storage, deletes the service, and restores it.
AnswerC

Azure AI Search allows updating encryption settings to use customer-managed keys on existing services without recreating the index.

Why this answer

Azure AI Search supports updating an existing service to use customer-managed keys by configuring Key Vault integration on the existing service instance.

166
MCQmedium

You are configuring an Azure AI Search index for use with Azure OpenAI On Your Data. The index contains technical documentation with complex nested tables. Which search configuration will yield the best retrieval performance for semantic queries?

A.Configure a pure OData filter index without vector embeddings.
B.Enable semantic search alongside vector and keyword search (hybrid search).
C.Use simple query syntax with exact phrase matching only.
D.Disable all vector fields and rely exclusively on traditional Lucene scoring.
AnswerB

Hybrid search combined with semantic ranking provides the highest retrieval accuracy for complex enterprise documents.

Why this answer

Enabling semantic search on the Azure AI Search index uses deep learning models to understand the intent and meaning of natural language queries, especially over complex document structures.

167
MCQeasy

When evaluating an AI agent, what does the 'Coherence' metric measure?

A.The cost of the API call.
B.The logical flow and clarity of the generated text.
C.The number of sources cited.
D.The speed of the response.
AnswerB

Coherence refers to the quality of the narrative structure.

Why this answer

Coherence measures how well the generated text flows and connects logical ideas together.

168
Multi-Selectmedium

You are designing a RAG solution using Azure AI Search and Azure OpenAI. Which TWO factors significantly influence the effectiveness of chunking documents during the ingestion phase? (Choose TWO)

Select 2 answers
A.Chunk size measured in tokens or characters
B.Azure OpenAI model temperature setting
C.Chunk overlap percentage between consecutive segments
D.Azure API Management timeout duration
E.Virtual network subnet IP address range
AnswersA, C

Chunk size determines how much specific detail versus broad context is captured in each vector.

Why this answer

Chunk size (token length) and chunk overlap are critical parameters that balance retaining context while ensuring chunks fit within embedding model token limits.

169
Multi-Selecteasy

You need to ensure data privacy and compliance when processing user feedback containing sensitive information. Which TWO types of sensitive data can Azure AI Language PII detection identify and redact? (Choose two.)

Select 2 answers
A.Weather forecast coordinates
B.Source code syntax errors
C.Phone numbers
D.Email addresses
E.CPU utilization metrics
AnswersC, D

Phone numbers are recognized and can be redacted by PII detection.

Why this answer

PII detection identifies personal data such as email addresses and phone numbers.

170
MCQmedium

You are developing a script to query a Custom Vision prediction endpoint. Which HTTP header must be included to authenticate your request using an API key?

A.Authorization: Bearer
B.X-Auth-Token
C.Vision-Api-Key
D.Prediction-Key
AnswerD

Custom Vision prediction endpoints require the Prediction-Key header (or Ocp-Apim-Subscription-Key).

Why this answer

Custom Vision prediction and training APIs authenticate using the `Prediction-Key` or `Training-Key` header respectively (or `Ocp-Apim-Subscription-Key`).

171
Multi-Selectmedium

Your organization requires that all Azure AI resources maintain rigorous logging and diagnostic tracking. Which TWO data categories can you collect using Azure Monitor diagnostic settings for Azure AI services? Each correct answer presents part of the solution.

Select 2 answers
A.Microsoft 365 Exchange mailbox audit logs
B.Request and response telemetry logs (such as Audit and Request logs)
C.Windows registry hives from underlying hypervisors
D.Platform performance metrics (such as Total Calls, Latency, and Failed Requests)
E.Active Directory domain controller SYSVOL replication status
AnswersB, D

Diagnostic settings capture detailed request telemetry and audit trails for AI services.

Why this answer

Diagnostic settings collect audit logs, request traces, and performance metrics.

172
MCQeasy

You are developing a .NET application that extracts bounding boxes of text from scanned documents using Azure AI Vision 4.0. Which asynchronous API operation should you call first to initiate the OCR process?

A.POST /computervision/imageanalysis:analyze?api-version=2024-02-01
B.POST /vision/v3.2/ocr
C.POST /vision/v2.1/recognizeText
D.POST /customvision/v3.3/Prediction
AnswerA

This is the correct endpoint and version for the unified Image Analysis 4.0 API which includes the Read feature.

Why this answer

Image Analysis 4.0 uses the Read API (Analyze Image 4.0 Read operation) for OCR, which is an asynchronous operation returning an Operation-Location header.

173
Multi-Selecthard

When configuring vector embeddings and search indexes in Azure AI Search for RAG, which THREE factors directly impact vector search performance and recall accuracy? (Choose THREE)

Select 3 answers
A.Embedding model dimension size and consistency with source generation
B.Azure App Service CPU architecture (x64 vs ARM)
C.HNSW (Hierarchical Navigable Small World) algorithm parameters such as m and efSearch
D.Choice of vector similarity metric (Cosine, Dot Product, or Euclidean)
E.Configuring Azure Front Door DNS CNAME records
AnswersA, C, D

Query and document vectors must use the exact same embedding model and dimensionality.

Why this answer

Embedding model dimensionality, vector similarity metric (e.g., cosine, dot product), and HNSW algorithm index parameters impact search performance and recall.

174
MCQeasy

You are setting up cost alerts for your Azure subscription, which contains several Azure AI services resources. You want to receive an email notification when your month-to-date spending reaches 80% of your $5,000 budget. Which feature should you configure?

A.Azure Cost Management budgets and alerts
B.Microsoft Defender for Cloud security alerts
C.Azure Monitor activity log alerts
D.Azure Advisor cost recommendations
AnswerA

Budgets in Azure Cost Management enable threshold alerts (e.g., 80%) to notify stakeholders via email.

Why this answer

Azure Cost Management budgets allow you to set spending thresholds and associate them with action groups for email notifications.

175
MCQhard

You are managing costs for a multi-tenant Azure AI project. You need to attribute usage costs specifically to the 'Research' department. What is the most effective approach?

A.Implement Azure Policy to enforce mandatory tags on all new resources.
B.Disable diagnostic logging for the research resources to reduce billing.
C.Create a separate resource group for each department.
D.Change the pricing tier of the resource to 'Free'.
AnswerA

Mandatory tagging ensures consistent cost reporting across all resources.

Why this answer

Tags are key-value pairs that help organize and track costs in Azure Cost Management.

176
MCQeasy

You are configuring a connection between Azure AI Foundry and an Azure OpenAI resource. Which credential type is recommended for secure, keyless authentication between services in Azure?

A.Account primary access key stored in application configuration
B.Hardcoded personal access tokens (PAT)
C.Anonymous public access with IP whitelisting
D.Managed Identity with Azure RBAC
AnswerD

Managed identity eliminates hardcoded API keys and is the security best practice across Azure AI services.

Why this answer

Managed identities (system-assigned or user-assigned) provide secure, keyless authentication between Azure resources without storing credentials in code or configuration files.

177
MCQhard

Your enterprise organization provisions multiple Azure AI Services accounts across different subscriptions. Security mandates that all API keys must be automatically rotated every 30 days without human intervention. Which Azure service configuration should you implement?

A.Microsoft Defender for Cloud JIT access policies
B.Azure Key Vault automated secret rotation using an Azure Function
C.Azure AI Services built-in automatic key rotation policy
D.Azure Monitor alert rules tied to Microsoft Entra ID
AnswerB

Azure Key Vault can trigger an Azure Function via Event Grid to programmatically regenerate and store new keys for Azure AI services on a schedule.

Why this answer

Azure Key Vault integrates with managed identities and Event Grid or Azure Automation to automate secret and API key rotation schedules.

178
MCQeasy

You are configuring an Azure AI Language resource in the Azure portal. You need to ensure that the text analysis solution supports custom text classification and named entity recognition. Which pricing tier should you select?

A.Premium (P1)
B.Standard (S)
C.Basic (B)
D.Free (F0)
AnswerB

The Standard tier supports all standard and custom text analysis capabilities.

Why this answer

Custom features such as custom text classification and custom named entity recognition require the Standard (S) pricing tier. The Free (F0) tier only supports base pre-built features with monthly limits.

179
MCQmedium

Your organization has strict compliance policies requiring all Azure AI Services endpoints to reject connections that do not use Transport Layer Security (TLS) version 1.3. Where can you enforce this minimum TLS version setting?

A.Within the Data residency tab of the Overview blade
B.In the Access control (IAM) settings under role assignments
C.In the Configuration blade of the Azure AI resource in the Azure Portal
D.In Application Insights under the Availability blade
AnswerC

The Configuration blade allows administrators to set the minimum TLS version required for incoming API calls.

Why this answer

The minimum TLS version for Azure AI Services can be enforced globally on the resource via the 'Configuration' blade in the Azure Portal or using Azure Policy. IAM, diagnostic settings, and Application Insights do not configure TLS versions.

180
Multi-Selecthard

You are deploying Custom Vision containers on-premises using Docker. Which THREE prerequisites and configuration steps are required? (Choose three.)

Select 3 answers
A.Ensure internet connectivity or periodic connection to Azure for billing metering.
B.Provide the Billing Endpoint URI and API Key via environment variables.
C.Accept the End-User License Agreement (EULA) by setting the EULA=accept environment variable.
D.Install the full Azure CLI inside every prediction container.
E.Grant root access to the Azure subscription billing portal from inside the container.
AnswersA, B, C

Containers report billing usage periodically back to Azure.

Why this answer

Custom Vision containers require an Azure AI Services or Custom Vision billing endpoint/key, acceptance of the EULA, and proper Docker runtime environment variables.

181
MCQmedium

You are configuring model fine-tuning for an Azure OpenAI model in Azure AI Foundry. You have prepared a JSONL training dataset. Which evaluation approach is recommended to verify that fine-tuning successfully improved model performance without causing catastrophic forgetting?

A.Inspect Azure Key Vault access audit logs
B.Review Azure Cost Management invoices for training hours
C.Run an evaluation dataset comparing baseline and fine-tuned models using Azure AI Foundry evaluation metrics
D.Check Azure Monitor CPU utilization percentages of the hosting cluster
AnswerC

Comparing baseline and fine-tuned models on a holdout test dataset validates improvements and detects regressions.

Why this answer

Evaluating the fine-tuned model against a holdout test dataset using automated metrics (like GPT-4 as a judge) compares baseline vs. fine-tuned performance.

182
MCQmedium

You are migrating a custom extraction solution from a Custom Template model to a Custom Neural model in Azure AI Document Intelligence. What is a key advantage of choosing a Custom Neural model over a Custom Template model?

A.Custom Neural models can automatically generalize across documents with varying visual layouts and structures.
B.Custom Neural models only support English language documents.
C.Custom Neural models require zero training documents.
D.Custom Neural models do not support table extraction.
AnswerA

Neural models use advanced deep learning to understand document semantics and layouts even when fields shift positions.

Why this answer

Custom Neural models support deep learning to handle documents with varying layouts and structures automatically, removing the need for strict template matching.

183
MCQeasy

You are setting up Azure AI Language. You want to authenticate your client application using Microsoft Entra ID (formerly Azure Active Directory) instead of using API subscription keys. Which role must be assigned to your application's service principal to allow it to call text analysis APIs?

A.Reader
B.Cognitive Services User
C.Storage Blob Data Contributor
D.Contributor
AnswerB

Cognitive Services User grants permissions to invoke data-plane inference APIs.

Why this answer

To call Azure AI Language APIs using Microsoft Entra ID, the security principal must be assigned the 'Cognitive Services User' role.

184
Multi-Selecthard

You are configuring an Azure AI Document Intelligence custom neural model training run. Which TWO prerequisites must be met regarding your training dataset? (Choose TWO)

Select 2 answers
A.Training documents must be stored in an Azure Blob Storage container with a SAS URI or managed identity access.
B.You must provide a minimum of five training documents for each defined field or form type.
C.All training documents must be converted to Microsoft Word (.docx) format prior to upload.
D.You must manually write a Python script to generate the labels.json file instead of using Document Intelligence Studio.
E.Training documents must all have identical pixel dimensions and DPI settings.
AnswersA, B

Training data for custom models must be accessible via Azure Blob Storage using a SAS token or managed identity.

Why this answer

Custom neural models require a minimum number of samples and specific file formats stored in an Azure Blob Storage container.

185
MCQeasy

Your application needs to scan user input for Personally Identifiable Information (PII) such as Social Security numbers, credit card numbers, and physical addresses before storing logs in a cloud database. Which Azure AI Language service feature should you invoke?

A.Sentiment analysis
B.Text summarization
C.Named Entity Recognition (NER) with PII detection
D.Key phrase extraction
AnswerC

PII detection via NER identifies and can redact sensitive personal data elements.

Why this answer

Named Entity Recognition (NER) for PII allows you to detect, categorize, and redact sensitive personal information in unstructured text.

186
MCQmedium

You are using Language Studio to build a custom named entity recognition (custom NER) project. You have tagged your training data and successfully trained your first model. You want to test the model interactively without writing code. Which tool in Language Studio should you use?

A.Deployment settings
B.Data labeling page
C.Test model pane
D.Model performance tab
AnswerC

The test pane in Language Studio allows interactive validation of trained models.

Why this answer

Language Studio provides a built-in 'Test' interface where you can input custom text and inspect the model's entity extractions interactively.

187
Multi-Selectmedium

Which THREE settings in the Azure AI Foundry playground can directly affect the 'creativity' of an agent's response?

Select 3 answers
A.Temperature.
B.The search index threshold.
C.Top-P (Nucleus sampling).
D.Frequency Penalty.
E.The system prompt length.
AnswersA, C, D

Temperature controls randomness.

Why this answer

Temperature, Top-P, and Frequency Penalty are the primary hyperparameters for controlling response output style.

188
MCQeasy

You are configuring an agent in Azure AI Foundry that needs to perform mathematical calculations accurately. Large language models often make arithmetic errors. How should you equip the agent to handle calculations reliably?

A.Provide a code interpreter tool so the model can write and execute Python code for calculations.
B.Set the model temperature to 0.0 and ask it to guess the result.
C.Increase the presence penalty to force new tokens.
D.Instruct the model in the system prompt to double-check its arithmetic mentally.
AnswerA

Executing Python code guarantees accurate mathematical computation.

Why this answer

Giving the agent a code interpreter or calculator tool allows it to execute calculations deterministically rather than relying on internal probabilistic math.

189
MCQmedium

A user asks a question that triggers a 'Blocked' response from Azure AI Content Safety, even though the query seems harmless. What is the best step to investigate this?

A.Change the system prompt to ignore safety.
B.Increase the model temperature.
C.Disable Content Safety for the deployment.
D.Check the Content Safety logs in the Azure AI Foundry portal.
AnswerD

Logs provide the justification for why a specific policy was triggered.

Why this answer

Viewing the logs in the Content Safety dashboard identifies which specific policy was triggered.

190
MCQeasy

You are deploying an Azure AI service resource in the Azure portal. Which option specifies the geographic location where your data will be processed and stored?

A.Tenant
B.Domain
C.Region
D.Zone
AnswerC

The Region setting determines the geographic location for resource deployment and data residency.

Why this answer

The 'Region' property determines the Azure datacenter location where resources are deployed and data is stored.

191
Multi-Selecteasy

Which THREE of the following are common Azure cost management tools that you should use to manage your AI solution budget?

Select 3 answers
A.Azure Budgets.
B.Azure Advisor cost recommendations.
C.Azure Cost Analysis.
D.Azure Active Directory.
E.Azure Container Registry.
AnswersA, B, C

Budgets allow you to set spending limits and receive notifications.

Why this answer

Budgets, Cost Analysis, and Advisor are the core tools for monitoring and optimizing spend.

192
MCQmedium

Your company is deploying an Azure OpenAI gpt-4 model. You need to implement model versioning so that when Microsoft releases a newer model version, you can test it on a small subset of user traffic before rolling it out to all users. Which feature should you configure?

A.Create multiple model deployments with different version tags and configure traffic splitting weights.
B.Enable dynamic scaling rules in Azure Kubernetes Service (AKS).
C.Configure Azure Traffic Manager with two regional endpoints.
D.Use Azure Application Gateway URL path-based routing.
AnswerA

Traffic splitting across deployments allows gradual percentage-based routing to new model versions.

Why this answer

Azure OpenAI model deployments support setting weight allocations or creating multiple deployments for canary testing and traffic splitting.

193
MCQhard

You are designing a disaster recovery strategy for an Azure AI Document Intelligence resource. The business requirement is a Recovery Point Objective (RPO) of zero and Recovery Time Objective (RTO) of less than 15 minutes. Which deployment strategy should you select?

A.Deploy two identical Azure AI resources in paired regions and use Azure Front Door with priority routing for automatic failover.
B.Configure geo-replication on a standard single-service resource using the Azure CLI without front-end routing.
C.Use Azure Site Recovery to replicate the Azure AI virtual machine host nodes directly.
D.Take daily backups of the trained custom models by exporting them to Azure Blob Storage in the same region.
AnswerA

Active-passive or active-active multi-region deployment with Azure Front Door ensures minimal RTO and RPO.

Why this answer

Deploying paired regions with geo-redundant storage (or multi-region active-active deployment using Azure Front Door) is required to meet aggressive RPO/RTO requirements. Standard local redundancy or manual JSON backups do not provide automated sub-15-minute failover.

194
Multi-Selectmedium

Which TWO techniques are effective for improving the performance of an Azure AI application?

Select 2 answers
A.Increasing the number of subscription keys.
B.Disabling all logging to reduce overhead.
C.Switching to the 'Free' tier.
D.Implementing caching for common query results.
E.Deploying the AI resource in a region closer to the end-users.
AnswersD, E

Caching reduces the need for expensive model inference calls.

Why this answer

Caching and regional proximity (or global distribution) directly improve performance.

195
MCQmedium

You are training a Custom Vision object detection model to identify specific manufacturing defects on a production line. You have uploaded 50 images and tagged the defects. When you attempt to train the model, the training fails. What is the most likely cause?

A.You must have at least 5 tags defined in the project.
B.Object detection requires a minimum of 15 tagged images per tag.
C.You must use multi-class classification instead of object detection for manufacturing defects.
D.Training images must be smaller than 256x256 pixels.
AnswerB

Custom Vision requires at least 15 images per tag for object detection projects.

Why this answer

Custom Vision object detection requires a minimum of 15 images per tag for the training set.

196
MCQhard

You are designing a high-throughput document processing pipeline using Azure AI Document Intelligence. To optimize latency and cost, you want to extract only the text and layout structures (paragraphs, lines, words, and tables) from a multi-page PDF without performing key-value extraction or using custom models. Which model ID should you specify in your API request?

A.prebuilt-layout
B.prebuilt-invoice
C.prebuilt-read
D.prebuilt-document
AnswerA

prebuilt-layout extracts text, tables, selection marks, and structure efficiently without key-value overhead.

Why this answer

The prebuilt layout model (prebuilt-layout) extracts document structures, tables, selection marks, and text without entity or custom key-value extraction.

197
Multi-Selecteasy

You are exploring Azure AI Language features. Which TWO tasks can be accomplished using prebuilt named entity recognition (NER)? (Choose two.)

Select 2 answers
A.Detecting the written language of a document
B.Extracting person names from text
C.Identifying geographic locations
D.Redacting credit card numbers for security
E.Classifying entire documents into custom business categories
AnswersB, C

Person names are recognized by prebuilt NER.

Why this answer

Prebuilt NER recognizes standard entities such as dates, times, people, and locations.

198
Multi-Selectmedium

You are monitoring an Azure OpenAI model deployed in Azure AI Foundry. Which TWO Azure diagnostic metrics or logs should you analyze to track operational health and performance? (Choose TWO)

Select 2 answers
A.Request duration (latency) in milliseconds
B.Azure ExpressRoute circuit bandwidth saturation
C.Prompt tokens and completion tokens consumed per request
D.Azure SQL Database deadlock frequency
E.Virtual machine disk IOPS consumption
AnswersA, C

Latency metrics help monitor model responsiveness and performance SLAs.

Why this answer

Request duration (latency) and token counts (prompt and completion tokens) are key operational metrics tracked in diagnostic logs.

199
Multi-Selectmedium

Which TWO of the following are valid ways to provide context to an agent in Prompt Flow?

Select 2 answers
A.Deploying the model in a different region
B.Passing conversation history as part of the messages list
C.Injecting retrieved documents into the prompt template
D.Increasing the model output length
E.Training a new model on the context
AnswersB, C

History provides multi-turn context.

Why this answer

Static prompt text and dynamic vector store retrieval are standard methods.

200
MCQeasy

You are creating an Azure AI Services resource using Azure CLI. Which command parameter specifies the pricing tier of the resource?

A.--sku
B.--tier
C.--pricing
D.--kind
AnswerA

The --sku parameter defines the pricing tier (e.g., F0, S0) when provisioning Azure AI services via Azure CLI.

Why this answer

The '--sku' parameter in Azure CLI commands (like 'az cognitiveservices account create') specifies the pricing tier.

201
MCQeasy

Your company is provisioning a new Azure AI Services multi-service resource. The security team mandates that all network traffic between the client applications and the Azure AI endpoint must remain entirely within the Microsoft backbone network without traversing the public internet. Which feature should you configure?

A.Cross-Origin Resource Sharing (CORS)
B.Service Principal Authentication
C.Azure Private Endpoint
D.Azure AI Custom Subdomain
AnswerC

Private Endpoint secures network traffic by bringing Azure AI Services inside your virtual network.

Why this answer

Azure Private Link enables you to access Azure AI Services via a private endpoint in your virtual network, ensuring traffic stays within the Microsoft backbone network. Public IP, DNS forwarding rules, and service tags do not inherently provide the required private network isolation.

202
Multi-Selectmedium

You are managing cost allocation for multiple Azure AI services across different departments. Which TWO Azure tools or features can help you analyze and attribute these costs accurately? Each correct answer presents part of the solution.

Select 2 answers
A.Azure Bastion session logs
B.Azure Active Directory enterprise user group memberships
C.Microsoft Defender for Cloud security recommendations
D.Azure resource tags combined with Cost Management cost analysis grouping
E.Azure Cost Management budgets and exported CSV cost reports
AnswersD, E

Tags allow categorizing costs by department, which Cost Management can group and report on.

Why this answer

Resource tags and Cost Management cost analysis are primary tools for cost attribution.

203
Multi-Selectmedium

You are troubleshooting a scenario where client applications are unable to connect to an Azure AI Language resource secured by a Private Endpoint. Which TWO troubleshooting steps should you perform? Each correct answer presents part of the solution.

Select 2 answers
A.Check that the client subnet can route to the private IP address assigned to the Private Endpoint.
B.Verify that the client application has been assigned the Storage Blob Data Contributor role.
C.Ensure that the Azure OpenAI model weights have been retrained.
D.Verify that the Private DNS zone is correctly linked to the client virtual network.
E.Check that public network access is set to enabled and unrestricted.
AnswersA, D

Routing rules must allow traffic from client subnets to the private endpoint IP.

Why this answer

Checking Private DNS zone configuration and verifying VNet peering or routing are key troubleshooting steps.

204
MCQmedium

You have an Azure AI Search service. You need to monitor the number of failed queries to identify potential indexing issues. Which tool should you use?

A.The 'Activity Log' in the Subscription blade.
B.The 'Security Center' recommendation dashboard.
C.The 'Metrics' blade in the Azure portal for the Search service.
D.The 'Resource Health' blade.
AnswerC

Metrics provide real-time tracking of Search query success and failure counts.

Why this answer

Azure Monitor metrics are the standard way to track service health and query performance.

205
Multi-Selectmedium

You are troubleshooting an Azure AI Language resource deployment. Which THREE error codes or HTTP status codes might indicate an issue with your API request structure or payload? (Choose three.)

Select 3 answers
A.404 Not Found
B.100 Continue
C.400 Bad Request
D.200 OK
E.422 Unprocessable Entity
AnswersA, C, E

404 indicates an invalid endpoint URI.

Why this answer

HTTP 400 (Bad Request) indicates malformed JSON or invalid parameters, HTTP 404 indicates an incorrect endpoint URL, and HTTP 422 indicates unprocessable data. HTTP 200 is success.

206
MCQeasy

You are reviewing the Azure Cost Management dashboard to analyze spending on Azure AI services. Which grouping option allows you to see costs broken down by specific resource names?

A.Resource type
B.Meter category
C.Resource
D.Location
AnswerC

Grouping by Resource displays individual cost line items for each deployed Azure AI resource.

Why this answer

Grouping by 'Resource' in Azure Cost Management displays costs associated with each individual resource instance.

207
MCQhard

You are designing an edge solution where a Custom Vision exported model runs inside a Linux container on an IoT Edge device without internet connectivity. How must you configure billing for the container?

A.Provide the container with a billing endpoint URL and API key, and ensure periodic connectivity to Azure to report billing metrics.
B.Use Custom Vision Training API keys for container runtime billing.
C.Purchase an Azure Storage account to store offline billing logs locally.
D.Deploy the container with no environment variables; billing is handled automatically by IoT Hub.
AnswerA

Cognitive services containers require billing endpoint credentials and periodic connectivity to report usage.

Why this answer

Custom Vision and cognitive services containers running disconnected or on edge devices require a billing endpoint and API key, but for fully offline scenarios without cloud connectivity, commitment tiers or specific edge billing configurations (such as cloud reporting when connected periodically) apply. However, containers generally require connection to report billing unless configured with specific volume mount billing mechanisms supported by Azure.

208
Multi-Selecthard

You are optimizing an application that calls Azure AI Vision Read API. Which THREE techniques help reduce latency and improve overall throughput? (Choose three.)

Select 3 answers
A.Send uncompressed RAW bitmap data to bypass image decoding overhead.
B.Disable SSL/TLS verification to speed up TCP handshakes.
C.Resize extremely large images to optimal dimensions while maintaining readability.
D.Implement proper polling intervals (e.g., waiting 1 second between checks) rather than aggressive busy-waiting.
E.Co-locate the application hosting and the Azure AI Vision resource in the same Azure region.
AnswersC, D, E

Reducing massive file sizes speeds up network transfer and processing.

Why this answer

Throughput and latency improvements include resizing excessively large images before upload, using asynchronous polling efficiently without tight busy-wait loops, and co-locating resources.

209
MCQhard

You are using Document Intelligence Studio to build a custom extraction model. You need to verify the model quality before deploying. Which metric is most critical?

A.Recall only
B.F1 Score
C.Precision only
D.Model ID
AnswerB

F1 score is the standard metric for balanced model evaluation.

Why this answer

The F1 score provides a balance between precision and recall, serving as the primary indicator of model quality.

210
MCQeasy

You are provisioning an Azure AI Language resource through the Azure CLI. You need to specify the resource kind parameter. Which value should you provide for a general text analysis resource?

A.LanguageService
B.TextUnderstanding
C.TextAnalytics
D.CognitiveServices
AnswerC

TextAnalytics is the correct resource kind parameter for Azure AI Language.

Why this answer

When provisioning via CLI or ARM templates, the kind for Azure AI Language is 'TextAnalytics'.

211
Multi-Selecthard

You are building an advanced agentic solution in Azure AI Foundry that integrates with external line-of-business systems. Which THREE architectural patterns or features should you incorporate to ensure robustness and safety? (Choose THREE)

Select 3 answers
A.Human-in-the-loop review for high-impact tool actions
B.Allowing the agent to execute any bash command directly on production servers without validation
C.Disabling all logging and monitoring to maximize execution speed
D.Exponential backoff and jitter retry logic for handling API rate limits (HTTP 429)
E.Structured Outputs with strict JSON Schema validation for tool payloads
AnswersA, D, E

Human approval prevents unintended data modifications or destructive API calls.

Why this answer

Rate limiting, human-in-the-loop approvals, and robust exception handling with retries are essential for robust agentic integrations.

212
MCQhard

You are troubleshooting a performance issue where requests to an Azure OpenAI model deployment experience intermittent HTTP 429 errors. You examine Azure Monitor metrics and see that the 'Processed Tokens' metric is well below your provisioned limit, but the 'Requests' metric spikes. What is the root cause of the error?

A.The model weights have become corrupted and require redeployment.
B.The request rate (Transactions Per Second) has exceeded the provisioned limit, even though token volume is within bounds.
C.The Azure subscription has run out of monthly financial budget.
D.The Private Endpoint DNS resolution has failed.
AnswerB

Azure OpenAI enforces both token-based and request-rate (Calls per Minute/Second) limits; exceeding either triggers HTTP 429.

Why this answer

HTTP 429 errors can be caused by exceeding Transactions Per Second (TPS) limits even if token quotas are not exhausted.

213
Multi-Selecthard

You are optimizing an agentic workflow in Azure AI Foundry. The agent needs to solve complex reasoning problems by breaking them down into distinct sub-tasks, executing them iteratively, and reflecting on intermediate results. Which THREE architectural frameworks or patterns support this behavior? (Choose three.)

Select 3 answers
A.Reflection pattern where the agent critiques its own intermediate outputs and refines them.
B.ReAct (Reason and Act) pattern combining thought generation with tool execution steps.
C.Static hardcoded IF-ELSE decision trees without model inference.
D.Single-turn stateless completion calls with temperature set to zero.
E.Plan-and-Solve prompt strategy where the agent creates a complete execution plan before stepping through tasks.
AnswersA, B, E

Self-reflection allows agents to catch errors and improve output quality iteratively.

Why this answer

Chain-of-thought, ReAct (Reason+Act), and plan-and-solve are foundational agentic reasoning patterns.

214
Multi-Selecthard

You are deploying an enterprise RAG application in Azure AI Foundry. You need to ensure data security and compliance across all components. Which THREE security practices should you implement? (Choose THREE)

Select 3 answers
A.Use Managed Identities with Azure RBAC for service-to-service authentication
B.Grant anonymous public network access on all AI service firewalls
C.Configure Azure Private Endpoints for Azure OpenAI and Azure AI Search
D.Store API keys in plain text configuration files committed to public GitHub repositories
E.Implement Customer-Managed Keys (CMK) via Azure Key Vault for data encryption at rest
AnswersA, C, E

Managed identities eliminate hardcoded credentials and enforce role-based access control.

Why this answer

Using private endpoints, managed identities, and customer-managed keys are foundational enterprise security practices for Azure AI and search services.

215
MCQmedium

You are building a RAG solution and observe that the model retrieves the correct documents but fails to use the information accurately. What is the most effective approach to improve performance?

A.Refine the system prompt instructions
B.Increase the vector index size
C.Use a smaller base model
D.Disable streaming response
AnswerA

Instructions guide the model on how to use context.

Why this answer

Iterative prompt engineering and improving the system message instructions are key to improving RAG performance.

216
Multi-Selecthard

Which TWO strategies are recommended to prevent 'Model Over-Reliance' or 'Automation Bias' in agents?

Select 2 answers
A.Include a clear disclaimer that the agent is an AI and may provide incorrect information.
B.Disable the ability for the agent to say 'I don't know'.
C.Provide citations and links back to original source documents.
D.Remove all human oversight from the process.
E.Use the highest available model temperature.
AnswersA, C

Transparency reduces user over-reliance.

Why this answer

Ensuring the user is informed and has a way to verify information prevents over-reliance on the agent's output.

217
Multi-Selectmedium

When designing an evaluation pipeline in Azure AI Foundry for generative AI applications, which TWO metrics are commonly used to assess safety and risk? (Choose TWO)

Select 2 answers
A.Database transaction log growth rate
B.Storage account geo-replication latency
C.Sexual content and violence safety metrics
D.Network bandwidth throughput in megabits per second
E.Hate speech and unfairness detection
AnswersC, E

Safety metrics measure the presence of harmful or explicit content in completions.

Why this answer

Hate speech, sexual content, violence, and self-harm detection metrics assess safety and risk in Azure AI Foundry evaluation pipelines.

218
MCQhard

You are developing an application that processes multilingual documents using Azure AI Language. You want to ensure that your API payload complies with the maximum document size limit per single request document in synchronous calls. What is the maximum character length allowed for a single document input?

A.5,120 characters
B.1,000,000 characters
C.50,000 characters
D.1,024 characters
AnswerA

The maximum character count per document for synchronous analysis is 5,120.

Why this answer

For the Azure AI Language REST API, the maximum character length for a single document input is 5,120 characters (as measured by StringInfo.LengthInTextElements).

219
MCQhard

You are training a Custom Template model using Azure AI Document Intelligence. Your training dataset contains documents with multiple pages where the tables span across page breaks. How should you structure your training labels to handle multi-page tables correctly?

A.Create a separate table label for each page fragment and concatenate them via post-processing code.
B.Exclude multi-page tables from the training set and rely solely on the prebuilt layout model for tables.
C.Label the table rows across pages using the same table index and consistent headers so the model recognizes the continuation.
D.Convert all multi-page documents into single extremely long virtual pages before labeling.
AnswerC

Consistent table indexing across pages in the labeling session allows the model to learn multi-page table structures accurately.

Why this answer

When training a custom template model, tables that span multiple pages must be labeled continuously with consistent table IDs across page breaks so the model learns to associate them as a single logical table.

220
MCQmedium

Your company uses Azure Policy to enforce governance across all subscriptions. A developer attempts to deploy an Azure OpenAI resource in a region that is prohibited by corporate policy. What happens during the deployment?

A.Azure automatically moves the resource to an allowed region.
B.The resource is deployed successfully, but an email alert is sent to the administrator.
C.The deployment succeeds and automatically applies a remediation task.
D.The deployment is blocked and returns an error explaining the policy violation.
AnswerD

Policy enforcement with 'Deny' rejects non-compliant resource creation requests immediately.

Why this answer

Azure Policy in 'Deny' mode blocks resource creation requests that violate policy rules.

221
MCQmedium

You are building an agent in Azure AI Foundry that requires access to a private internal database. Which component should you configure to ensure the agent retrieves data securely without exposing the database endpoint publicly?

A.Azure Private Link and Managed Virtual Network
B.Public IP allowlisting
C.Entra ID Application Proxy
D.Shared Access Signatures (SAS)
AnswerA

Managed Virtual Networks and Private Endpoints provide secure, private connectivity.

Why this answer

Managed Virtual Networks and Private Endpoints in Azure AI Foundry ensure that traffic between your agent and data sources remains within the Microsoft backbone network.

222
MCQhard

You are migrating an existing solution from the legacy Text Analytics API v3.1 to the current Azure AI Language REST API. You notice changes in the JSON request body structure. Where should the document text array now be nested in the payload for a sentiment analysis request?

A.Inside the HTTP header as a base64 encoded string.
B.Directly under the root object as a 'documents' array.
C.Nested inside a 'parameters' object.
D.Nested inside an 'analysisInput' object under the root.
AnswerD

The current Azure AI Language API nests documents inside an 'analysisInput' object.

Why this answer

In the Azure AI Language API (such as version 2022-05-01 and later), documents are wrapped inside an analysis input object, specifically under 'analysisInput': { 'documents': [...] }.

223
MCQmedium

You want to add a new tool to an agent. Which step is required to link the tool to the model?

A.Hardcode the tool logic in the system prompt.
B.Add the tool code to the global Python environment.
C.Register the tool in the agent configuration within Azure AI Foundry.
D.Update the content safety policy.
AnswerC

Tools must be registered and mapped for the model to recognize their availability.

Why this answer

The tool must be defined as an 'Available Tool' in the agent's configuration so the model knows it can call it.

224
Multi-Selecthard

Your company's compliance mandate requires that all data processed by Azure AI Speech and Document Intelligence services remain encrypted using keys managed entirely by your organization. Which THREE configuration steps are required? Each correct answer presents part of the solution.

Select 3 answers
A.Assign appropriate permissions for the AI service managed identity to access the Key Vault key.
B.Store the private key as plain text in the application source code.
C.Create an Azure Key Vault and enable soft-delete and purge protection.
D.Configure the customer-managed key (CMK) identifier on the Azure AI services account properties.
E.Disable all diagnostic logging across all Azure subscriptions.
AnswersA, C, D

The AI resource's identity must have read/wrap/unwrap permissions on the key.

Why this answer

CMK setup requires Key Vault creation, enabling soft-delete/purge protection, and linking the key to the AI service.

225
Multi-Selectmedium

Which TWO strategies are recommended for handling high-volume traffic to an Azure AI endpoint?

Select 2 answers
A.Enabling autoscaling on the endpoint.
B.Limiting access to a single user.
C.Hardcoding the IP address of the compute node.
D.Using only the 'Free' pricing tier.
E.Implementing a global load balancer (e.g., Front Door).
AnswersA, E

Autoscaling allows the service to handle spikes automatically.

Why this answer

Load balancing and autoscaling are the key patterns for managing high traffic.

Page 2

Page 3 of 7

Page 4

All pages