AI-102 Azure AI Engineer Associate Practice Questions

1008 exam-accurate questions with explanations

info

Free Sample Questions

Showing 10 of 1008 questions. Get full access to all questions, detailed explanations, and study materials.

Plan and manage an Azure AI solution

Blue Yonder plans to deploy an Azure AI container at an edge site with intermittent internet. What should be considered before choosing this design?

A Containers always support every Azure AI feature

B Containers require Azure Kubernetes Service only

C The container must still meet licensing, billing, and connectivity requirements for the selected service check_circle

D Containers eliminate all Azure billing requirements

lightbulb

Explanation

Azure AI containers can support edge scenarios, but they have service-specific licensing, metering, and connectivity requirements. The best wrong choice is tempting because AKS can run containers, but it is not the only possible runtime, but it does not meet the stated requirement.

Plan and manage an Azure AI solution

Woodgrove Bank, a travel company, needs one Azure AI resource that can call multiple Azure AI services from a single endpoint and key. The solution must support role-based access control. What should you create?

A A separate single-service resource for each capability

B A multi-service Azure AI services resource check_circle

C An Azure Machine Learning workspace only

D An Azure Storage account with a system-assigned identity

lightbulb

Explanation

A multi-service Azure AI services resource provides a unified endpoint/key for several Azure AI services. The best wrong choice is tempting because single-service resources can be useful for isolation, but they do not provide one shared endpoint for multiple Azure AI services, but it does not meet the stated requirement.

Plan and manage an Azure AI solution

You must reduce blast radius between two unrelated production AI workloads that require different access policies and quotas. What should you provision? The implementation must follow current Microsoft guidance.

A One shared key in all applications

B A single prompt template for both workloads

C Separate Azure AI resources with separate identities, keys, and RBAC scopes check_circle

D A single knowledge store table

lightbulb

Explanation

Separate resources improve isolation for security, quota, and lifecycle management. The best wrong choice is tempting because a shared key increases blast radius and makes access harder to separate, but it does not meet the stated requirement.

Plan and manage an Azure AI solution

A team needs to track failed requests, latency, and quota usage for an Azure AI resource. Which Azure capability should they enable first? The solution will be deployed to production next quarter.

A Azure Bastion on the resource group

B Diagnostic settings to send metrics and logs to Azure Monitor or Log Analytics check_circle

C A lifecycle management policy on storage

D A custom vision training project

lightbulb

Explanation

Diagnostic settings and Azure Monitor/Log Analytics are the standard way to collect operational telemetry. The best wrong choice is tempting because Bastion is for secure VM access, not AI resource monitoring, but it does not meet the stated requirement.

Plan and manage an Azure AI solution

A company wants a governance framework for AI solutions. Which activity best aligns with Responsible AI before production release? The exam question focuses on the best Azure-native choice.

A Use only the largest available model

B Define risk controls, human review paths, monitoring, and content safety policies check_circle

C Increase temperature to make answers more creative

D Disable logging to avoid operational overhead

lightbulb

Explanation

Responsible AI governance includes risk management, oversight, monitoring, and safety controls. The best wrong choice is tempting because larger models can improve capability, but model size alone is not a governance framework, but it does not meet the stated requirement.

Plan and manage an Azure AI solution

You need to prevent harmful user prompts from reaching a generative AI application and detect prompt injection attempts. Which capability should you include? The team is choosing the first service or feature to implement.

A Prompt shields and content safety checks check_circle

B Only speech-to-text

C A custom translation model

D A knowledge store projection

lightbulb

Explanation

Prompt shields and content safety controls are used to reduce prompt attacks and harmful content. The best wrong choice is tempting because speech-to-text changes audio into text but does not moderate prompt attacks, but it does not meet the stated requirement.

Plan and manage an Azure AI solution

A solution architect must choose between Azure AI Vision, Azure AI Language, Azure AI Speech, Azure AI Search, and Azure Document Intelligence. The requirement is to extract fields from invoices and receipts. Which service is most appropriate?

A Azure AI Vision object detection

B Azure AI Speech transcription

C Azure AI Language sentiment analysis

D Azure AI Document Intelligence in Foundry Tools check_circle

lightbulb

Explanation

Document Intelligence is designed for form and document field extraction, including prebuilt invoice/receipt scenarios. The best wrong choice is tempting because Vision can perform OCR, but it does not provide the same document field extraction workflow, but it does not meet the stated requirement.

Plan and manage an Azure AI solution

An application must choose a default endpoint when using an Azure AI resource from SDK code. What information is generally required along with authentication?

A The service endpoint URI for the Azure AI resource check_circle

B The local file path of the model card

C The Azure subscription display name only

D The user interface language of the Azure portal

lightbulb

Explanation

SDK clients typically require the service endpoint plus a credential. The best wrong choice is tempting because the subscription display name does not tell the SDK where to send requests, but it does not meet the stated requirement.

Plan and manage an Azure AI solution

A regulated workload requires only approved users to call an Azure AI resource. Which control is the strongest fit?

A Use a public endpoint without authentication

B Assign least-privilege Azure RBAC roles and use Microsoft Entra authentication where supported check_circle

C Rely only on file naming conventions

D Share an admin key with all developers

lightbulb

Explanation

RBAC with Microsoft Entra authentication supports least privilege and auditable access. The best wrong choice is tempting because sharing an admin key weakens auditability and least privilege, but it does not meet the stated requirement.

Plan and manage an Azure AI solution

Your CI/CD pipeline must deploy AI resources consistently across dev, test, and prod. Which approach best supports repeatable deployment? The implementation must follow current Microsoft guidance.

A Create each resource manually in the portal

B Export the browser session cookies

C Copy keys from dev into prod

D Use infrastructure as code such as Bicep or ARM templates with environment-specific parameters check_circle

lightbulb

Explanation

Infrastructure as code supports repeatable, reviewable deployments across environments. The best wrong choice is tempting because manual portal creation is possible but error-prone and not ideal for CI/CD, but it does not meet the stated requirement.

Get all 1008 questions

Full access includes all questions, detailed explanations, PDF downloads, and timed mock exams.

quiz Start Free Practice download Free PDF