This post/practice exam is a part of the AI-200: Developing AI Cloud Solutions on Azure Exam Prep Hub.
Section 1 — Develop Containerized Solutions on Azure
Questions 1–7
Question 1 — Single Answer
You are developing an AI-powered document-processing application. The application is packaged as a Docker image and must be deployed to Azure App Service.
The development team wants App Service to retrieve the image from Azure Container Registry (ACR).
Which configuration is required?
A. Configure the App Service container settings to use the ACR image and provide the required registry authentication
B. Create an Azure Service Bus subscription between ACR and App Service
C. Configure an Event Grid subscription that automatically converts the image into an App Service package
D. Store the Docker image in Azure Blob Storage and configure App Service to mount the blob
Answer: A
Explanation: Azure App Service can run containerized applications using images stored in a container registry such as ACR. The App Service must be configured with the appropriate container image and registry authentication. A Service Bus subscription or Event Grid subscription is not required simply to deploy the image.
Question 2 — Single Answer
A development team wants to automatically build a container image whenever source code is committed. They want the container image to be built directly within Azure without maintaining a dedicated build server.
Which Azure Container Registry capability should you use?
A. ACR replication
B. Azure Container Registry Tasks
C. Azure Container Apps revisions
D. AKS Jobs
Answer: B
Explanation: Azure Container Registry Tasks can automate container image builds and related tasks. They can build images in Azure and can be integrated with source-control or registry workflows. This eliminates the need to maintain a dedicated build environment for the image-building process.
Question 3 — Scenario-Based, Single Answer
A company deploys an AI inference API to Azure Container Apps.
The application receives normal traffic throughout the day, but occasionally receives bursts of messages from an Azure messaging system. The company wants the number of application instances to automatically increase based on the number of pending messages.
Which capability should you use?
A. Container Apps revision traffic splitting
B. Azure App Service autoscale rules based on CPU
C. Kubernetes Event-driven Autoscaling (KEDA)
D. Azure Container Registry Tasks
Answer: C
Explanation: Azure Container Apps supports event-driven scaling through KEDA. KEDA allows scaling decisions to be based on external event sources, such as message queues, rather than relying solely on CPU or memory utilization.
Question 4 — Multiple Answers
You are troubleshooting an application running in Azure Container Apps.
Which two actions can help diagnose application problems?
A. Inspect application logs
B. Inspect container/application events
C. Change the Cosmos DB consistency level
D. Create a Service Bus topic
Answers: A, B
Explanation: Container Apps troubleshooting can involve examining logs and events associated with the application and its containers. Changing Cosmos DB consistency or creating a Service Bus topic does not directly diagnose a Container Apps deployment problem.
Question 5 — Matching
Match each Azure container technology or feature with the most appropriate description.
| Technology | Description |
|---|---|
| 1. Azure Container Registry | A. Serverless container hosting with application revisions |
| 2. Azure Container Apps | B. Managed container image registry |
| 3. AKS | C. Managed Kubernetes orchestration |
| 4. KEDA | D. Event-driven autoscaling |
Answer:
- 1 → B
- 2 → A
- 3 → C
- 4 → D
Explanation: ACR provides container image storage and management; Container Apps provides managed application hosting for containers; AKS provides managed Kubernetes; and KEDA provides event-driven autoscaling capabilities used by Container Apps.
Question 6 — Single Answer
You have deployed an application to AKS using a Kubernetes manifest.
The application starts successfully, but users cannot reach it through the expected endpoint.
Which approach should you take first to investigate the problem?
A. Rebuild the container image using ACR Tasks
B. Increase the Cosmos DB RU/s limit
C. Inspect Kubernetes resources, events, logs, and connectivity configuration
D. Create an Event Grid custom topic
Answer: C
Explanation: Microsoft specifically identifies monitoring and troubleshooting AKS applications through logs, events, and end-to-end connectivity as part of the AI-200 skill set. A deployment can succeed while networking, service, ingress, or application configuration prevents clients from reaching the application.
Question 7 — Single Answer
A new version of a containerized AI API is deployed to Azure Container Apps. The team wants to test the new version without immediately sending all production traffic to it.
Which Container Apps capability is most appropriate?
A. Container Apps revisions
B. ACR Tasks
C. Kubernetes DaemonSets
D. Azure Functions bindings
Answer: A
Explanation: Azure Container Apps supports revisions, allowing different versions of an application to be deployed and managed independently. Traffic can then be managed between revisions, supporting controlled rollout scenarios.
Section 2 — Develop AI Solutions Using Azure Data Management Services
Questions 8–15
Question 8 — Single Answer
An AI application uses Azure Cosmos DB for NoSQL to store documents and their embeddings.
The application needs to retrieve documents whose embeddings are semantically similar to a query embedding.
Which capability should the application use?
A. Change feed
B. Vector similarity search
C. Azure Functions timer trigger
D. Event Grid retry policy
Answer: B
Explanation: Azure Cosmos DB for NoSQL supports storing embeddings and performing vector similarity search. This capability is useful for semantic retrieval and AI/RAG scenarios.
Question 9 — Scenario-Based, Single Answer
An application frequently queries a Cosmos DB for NoSQL container using a property called category.
The query workload has grown significantly, and the development team wants to reduce unnecessary RU consumption by ensuring frequently queried properties are indexed appropriately.
What should the team investigate?
A. Cosmos DB indexing policy
B. Azure Service Bus lock duration
C. Container Apps revision settings
D. Azure Key Vault rotation policy
Answer: A
Explanation: Cosmos DB indexing policies affect how queries are executed and can influence RU consumption and query performance. Designing an appropriate indexing policy is an important part of optimizing Cosmos DB workloads.
Question 10 — Multiple Answers
An AI application uses Azure Cosmos DB for NoSQL.
Which two capabilities are directly relevant to implementing AI retrieval workflows?
A. Storing embeddings
B. Performing vector similarity searches
C. Creating App Service deployment slots
D. Configuring Service Bus dead-letter queues
Answers: A, B
Explanation: Cosmos DB for NoSQL can store embeddings and perform vector similarity searches. These capabilities support semantic retrieval and RAG-style AI applications.
Question 11 — Single Answer
A document-processing application needs to react whenever new or updated items are written to an Azure Cosmos DB for NoSQL container.
Which Cosmos DB capability is designed for this purpose?
A. Analytical store
B. Change feed
C. Vector index
D. Consistency policy
Answer: B
Explanation: The Cosmos DB change feed provides an ordered record of changes to items in a container and can be used to detect and process new or updated items. The AI-200 study guide explicitly includes implementing a change feed processor.
Question 12 — Fill in the Blank
Complete the statement:
Azure Database for PostgreSQL can support vector similarity search for AI workloads by using the __________ extension.
Answer: pgvector
Explanation: PostgreSQL AI workloads can use the pgvector extension to store and search vector embeddings. Microsoft specifically identifies Azure Database for PostgreSQL with pgvector as an AI data service covered by the course and exam.
Question 13 — Scenario-Based, Single Answer
A company is implementing a RAG application using Azure Database for PostgreSQL.
Each document chunk contains:
- an embedding
- document text
- a tenant ID
- a document type
The application must retrieve semantically similar chunks but only from the current tenant.
What should the application implement?
A. Vector similarity search combined with metadata filtering
B. A Service Bus subscription filtered by tenant ID
C. An Azure Function timer trigger
D. A Redis expiration policy
Answer: A
Explanation: Vector similarity search retrieves semantically related records, while metadata filtering restricts the result set to relevant records such as a particular tenant. The AI-200 study guide specifically calls out semantic retrieval and RAG patterns using metadata filters with PostgreSQL.
Question 14 — Single Answer
An AI application performs expensive vector searches against Azure Database for PostgreSQL.
The team wants to reduce vector-search latency and unnecessary computational overhead.
Which area should they investigate?
A. Azure Event Grid retry policies
B. pgvector indexing and query optimization
C. App Configuration feature flags
D. Service Bus dead-letter queues
Answer: B
Explanation: PostgreSQL vector workloads can be optimized through appropriate indexing strategies and pgvector configuration. Microsoft specifically includes optimizing vector search and reducing pgvector compute overhead in the AI-200 objectives.
Question 15 — Matching
Match each Azure data technology with the most appropriate use case.
| Technology | Use case |
|---|---|
| 1. Cosmos DB for NoSQL | A. Distributed document storage and vector search |
| 2. Azure Database for PostgreSQL | B. Relational data and pgvector-based AI workloads |
| 3. Azure Managed Redis | C. Low-latency caching and vector storage/search |
| 4. Cosmos DB change feed | D. Detecting new or updated Cosmos DB items |
Answer:
- 1 → A
- 2 → B
- 3 → C
- 4 → D
Explanation: These are distinct but complementary AI data patterns. Cosmos DB supports NoSQL document workloads and vector search; PostgreSQL supports relational workloads and pgvector; Azure Managed Redis supports caching and vector capabilities; and Cosmos DB change feed supports processing item changes.
Section 3 — Connect to and Consume Azure Services
Questions 16–23
Question 16 — Single Answer
An AI application submits long-running document-processing requests.
The API should return quickly rather than waiting for processing to finish. A backend worker should process the requests asynchronously.
Which Azure service is most appropriate for queuing the work?
A. Azure Key Vault
B. Azure Service Bus
C. Azure App Configuration
D. Azure Container Registry
Answer: B
Explanation: Azure Service Bus provides messaging capabilities appropriate for decoupling producers from consumers and processing backend operations asynchronously. The AI-200 objectives specifically include queuing and processing backend operations using Service Bus.
Question 17 — Scenario-Based, Single Answer
An AI application uses Azure Service Bus.
Messages that repeatedly fail processing should be isolated so that they do not continue to interfere with normal message processing.
What should the application use?
A. A dead-letter queue
B. An Event Grid custom topic
C. An App Configuration key
D. A Cosmos DB change feed
Answer: A
Explanation: Azure Service Bus provides dead-letter queues for messages that cannot be successfully processed or that meet specified dead-lettering conditions. This allows problematic messages to be isolated for inspection or later processing.
Question 18 — Multiple Answers
You are designing an AI workflow using Azure Service Bus.
Which two capabilities are directly supported by Service Bus and relevant to the AI-200 objectives?
A. Topics and subscriptions
B. Dead-letter queues
C. Vector similarity search
D. Container image versioning
Answers: A, B
Explanation: The AI-200 study guide explicitly includes Service Bus messages, topics, subscriptions, and dead-letter queue handling. Vector search and container image versioning are handled by other Azure services.
Question 19 — Single Answer
A company wants an application to react whenever a new image is added to an Azure Storage account.
The workflow should be event-driven rather than continuously polling the storage account.
Which Azure service should be used to route the event?
A. Azure Event Grid
B. Azure Key Vault
C. Azure Managed Redis
D. Azure Container Registry Tasks
Answer: A
Explanation: Event Grid is designed for event-driven architectures. It can route events from Azure resources to handlers and supports event filtering and retry behavior.
Question 20 — Scenario-Based, Single Answer
An AI application needs a lightweight HTTP API that performs a small amount of processing and then returns a response.
The API does not require a continuously running server.
Which Azure service is most appropriate?
A. Azure Kubernetes Service
B. Azure Functions
C. Azure Container Registry
D. Azure Service Bus
Answer: B
Explanation: Azure Functions provides serverless execution and supports HTTP-triggered functions that can implement APIs without requiring developers to manage a continuously running server infrastructure. The AI-200 objectives include building serverless APIs using triggers and bindings.
Question 21 — Single Answer
You are developing an Azure Function that should execute whenever a message arrives in a supported messaging system.
Which Azure Functions capability determines what causes the function to execute?
A. Binding
B. Trigger
C. Revision
D. Indexing policy
Answer: B
Explanation: A Function trigger defines the event that causes a function to execute. Bindings provide a declarative way to connect a function to input and output data or services.
Question 22 — Multiple Answers
Which two statements correctly describe Azure Functions concepts?
A. A trigger determines when a function executes.
B. Bindings can simplify interaction with external data or services.
C. A trigger is primarily used to create a PostgreSQL vector index.
D. Bindings replace all Azure authentication mechanisms.
Answers: A, B
Explanation: Triggers define function execution events, while bindings provide connections to data and services. They do not create database indexes or eliminate the need for authentication and authorization.
Question 23 — Scenario-Based, Single Answer
An organization publishes custom business events for its AI workflow. Several downstream applications should independently receive the events.
The organization also needs to apply event filters so that each subscriber receives only the events relevant to it.
Which Azure service is the best fit?
A. Azure Service Bus queue
B. Azure Event Grid
C. Azure Key Vault
D. Azure App Service
Answer: B
Explanation: Event Grid is designed for event-driven architectures and supports custom events and event filtering. It is appropriate when publishers emit events and multiple subscribers independently react to those events.
Section 4 — Secure, Monitor, and Troubleshoot Azure Solutions
Questions 24–30
Question 24 — Single Answer
An AI application requires an API key to access an external AI service.
The development team currently stores the key directly in the application’s source code.
What is the best Azure-native solution?
A. Store the key in Azure Container Registry
B. Store the key in Azure App Configuration as plain text
C. Store the key in Azure Key Vault
D. Store the key in an Azure Service Bus message
Answer: C
Explanation: Azure Key Vault is designed to securely store and retrieve secrets such as API keys, passwords, and other sensitive configuration information. Hard-coding secrets in source code should be avoided.
Question 25 — Scenario-Based, Single Answer
A company wants to change application configuration values without rebuilding and redeploying the application.
The configuration includes feature flags and non-secret application settings.
Which service should be used?
A. Azure App Configuration
B. Azure Key Vault
C. Azure Container Registry
D. Azure Service Bus
Answer: A
Explanation: Azure App Configuration provides centralized management of application settings and configuration. Key Vault is primarily intended for secrets and other sensitive values.
Question 26 — Multiple Answers
A company is designing secure configuration management for an AI application.
Which two approaches are appropriate?
A. Store sensitive secrets in Azure Key Vault.
B. Store application configuration information in Azure App Configuration.
C. Store API secrets directly in application source code.
D. Put database passwords into Event Grid event payloads.
Answers: A, B
Explanation: Key Vault is designed for secrets, while App Configuration provides centralized application configuration management. Secrets should not be embedded in source code or unnecessarily exposed in event payloads.
Question 27 — Single Answer
An AI application consists of multiple distributed services.
A request enters an API, invokes a backend service, calls a database, and then invokes another service.
The development team needs to follow the request across these components to identify where latency is occurring.
Which technology should they use?
A. Azure Container Registry Tasks
B. OpenTelemetry
C. Cosmos DB indexing
D. Service Bus dead-lettering
Answer: B
Explanation: OpenTelemetry provides standardized application instrumentation for collecting telemetry such as traces, metrics, and related diagnostic information. Distributed tracing can help follow a request across multiple services.
Question 28 — Scenario-Based, Single Answer
You need to analyze application logs stored in Azure Monitor Logs.
You want to find all requests where the duration was greater than 2 seconds and return the most recent results first.
Which technology should you use?
A. Dockerfile
B. KQL
C. pgvector
D. KEDA
Answer: B
Explanation: Kusto Query Language (KQL) is used to query and analyze data in Azure Monitor Logs and other Azure data platforms. The AI-200 study guide explicitly includes writing KQL queries to analyze logs and metrics.
Question 29 — Matching
Match each technology with its primary purpose.
| Technology | Purpose |
|---|---|
| 1. Azure Key Vault | A. Centralized application configuration |
| 2. Azure App Configuration | B. Secure secret management |
| 3. OpenTelemetry | C. Application telemetry and distributed tracing |
| 4. KQL | D. Query and analyze telemetry/log data |
Answer:
- 1 → B
- 2 → A
- 3 → C
- 4 → D
Explanation: Key Vault handles secrets, App Configuration manages application settings, OpenTelemetry provides instrumentation and telemetry, and KQL is used to query and analyze telemetry and log data.
Question 30 — Scenario-Based, Single Answer
An AI application is experiencing intermittent performance problems.
The application consists of several microservices running in Azure. Users report that some requests take several seconds to complete, but CPU utilization on the individual services does not consistently appear high.
The development team wants to determine which downstream service is contributing to the latency for individual requests.
Which approach is most appropriate?
A. Enable distributed tracing with OpenTelemetry
B. Increase the Cosmos DB consistency level
C. Create an additional Azure Container Registry
D. Configure a Service Bus dead-letter queue
Answer: A
Explanation: Distributed tracing is particularly valuable when an application spans multiple services. OpenTelemetry instrumentation can provide trace information that helps developers follow a request across service boundaries and identify where latency is introduced.
Go to the AI-200 Exam Prep Hub main page
The Data Community 