Microsoft Certified: Azure AI Cloud Developer Associate (AI‑200) Practice Exams

About the Azure AI-200 exam

Exam at a glance

AI-200, Developing AI Cloud Solutions on Azure, is Microsoft's associate certification for developers who build the back end that an AI solution runs on. Passing it earns the Microsoft Certified: Azure AI Cloud Developer Associate credential. It is the successor to AZ-204, the Azure Developer Associate exam that retires on July 31, 2026. The name says AI, but the blueprint is a developer blueprint: container hosting, data services with vector search, messaging and eventing, and the security, tracing, and troubleshooting that keep the whole thing alive in production. The exam runs 120 minutes, the passing score is 700 out of 1000, and it is currently offered in English only.

Domain weighting

• Develop containerized solutions on Azure: 20–25%
• Develop AI solutions by using Azure data management services: 25–30%
• Connect to and consume Azure services: 20–25%
• Secure, monitor, and troubleshoot Azure solutions: 20–25%

Who this exam is for

AI-200 targets developers who contribute to all phases of implementing AI solutions on Azure with an emphasis on back-end services and components: requirements gathering, design, development, deployment, security, and monitoring. Python is the stated implementation language. Questions are scenario-based: a short stem names a concrete Azure workload and two or three simultaneous constraints, and you pick the service, feature, mode, trigger, index, consistency level, or configuration that satisfies all of them.

Prerequisites

Microsoft's study guide lists no prerequisite exam. It does expect proficiency in the Azure SDKs and the third-party SDKs used on Azure, Azure data management services, Azure monitoring and troubleshooting, Azure messaging and eventing, vector databases, Python programming, and implementing containerized applications on Azure. If you already hold AZ-204, much of the container, messaging, Functions, Key Vault, and monitoring material will be familiar, and your extra study time goes to the vector-search and data-store depth that AI-200 adds.

Why take this certification

• It is the AZ-204 successor. The Azure Developer Associate exam retires on July 31, 2026, and AI-200 is the exam that carries the Azure back-end developer role forward.
• Back-end depth, not model theory. The blueprint asks you to build, store, version, deploy, configure, connect, optimize, secure, monitor, and troubleshoot — the plumbing around a model, not the model itself.
• Vector data where teams actually put it. Vector similarity search appears in three places on the blueprint: Azure Cosmos DB for NoSQL, pgvector on Azure Database for PostgreSQL, and vector indexing in Azure Managed Redis.
• Builder altitude, not trivia. Questions reward implementation judgment — the right trigger, the right index, the right consistency level, the right least-privilege credential — over memorized definitions.

What you'll learn in the AI-200 exam

AI-200 is hands-on and scenario-driven. Most questions describe a short implementation task where two or three Azure options all look plausible, and the better answer follows a single instinct that runs through the whole exam: when two answers both work, take the one the platform performs for you rather than the one you hand-code, and reject each near-miss on the one stated constraint it violates. The exam's favourite move is a scenario written to read like an application bug when the fix is a platform setting.

Core services and tools you'll be tested on

• Develop containerized solutions on Azure: building, storing, versioning, and managing images in Azure Container Registry; building and running images with ACR Tasks; deploying containers to Azure App Service, including supplying environment variables and secrets; deploying to Azure Container Apps with environment configuration and revision management; event-driven scaling with Kubernetes Event-driven Autoscaling (KEDA) in Container Apps; deploying and managing applications on Azure Kubernetes Service with manifest files; and troubleshooting AKS and Container Apps by inspecting logs, events, and end-to-end connectivity.
• Develop AI solutions by using Azure data management services: connecting to Azure Cosmos DB for NoSQL with the SDK and running queries; tuning Request Unit consumption with indexing policies and consistency levels; storing and retrieving embeddings and running vector similarity search; implementing a change feed processor; connecting to and querying Azure Database for PostgreSQL with SDKs, modelling schemas and indexes, sizing compute, memory, and storage for vector workloads, reducing pgvector compute overhead, running retrieval-augmented generation with metadata filters, and optimizing connections for throughput and latency; plus Azure Managed Redis caching, expiration, and invalidation, and Redis vector indexing for similarity search.
• Connect to and consume Azure services: queuing and processing back-end operations with Azure Service Bus, including messages, topics, subscriptions, and dead-letter queue handling; event-driven workflows with Azure Event Grid, including filters, custom events, and retries; and Azure Functions, from building serverless APIs with triggers and bindings to configuring and deploying function apps.
• Secure, monitor, and troubleshoot Azure solutions: securing secrets with Azure Key Vault, including rotation and retrieval; storing and retrieving application configuration with Azure App Configuration; tracing distributed systems with the OpenTelemetry SDKs; and writing KQL queries to analyze logs and metrics.

Implementation judgment patterns you'll need to recognize

• Pinning what actually runs: a SHA-256 manifest digest versus a mutable tag versus latest, and which ACR Tasks trigger — base-image update, source commit, schedule, or manual — matches the rebuild the scenario describes.
• Deciding who holds the work and who initiates delivery: one worker that must take each item and prove it finished is Service Bus in PeekLock, several independent handlers that each need to be told the same thing happened is Event Grid.
• Picking the scaling story: a KEDA queue-length scaler with a minimum of zero replicas versus a fixed minimum, HTTP-concurrency scaling, or an always-on App Service plan, given a cost and latency target.
• Matching the index to the query: pgvector HNSW or IVFFlat versus a B-tree, a RediSearch vector field with a KNN query versus a key scan, and a Cosmos DB indexing policy that stops paying Request Units for paths nobody queries.
• Choosing the consistency level the scenario actually needs — read-your-own-writes across nodes points at Session, not automatically at Strong — and knowing where the session token is emitted and supplied.
• Getting configuration and secrets in the managed way: Key Vault references and managed identity over stored keys or values pasted into settings, App Configuration labels versus content types versus key prefixes, and rotating without a redeploy.
• Making a running system observable before you need it: the span survives only because something instrumented the code and a collection setting was in place before the event happened, which is why the fix is often a grant, a reference form, a trigger, or a collection setting rather than a code change.

How the practice exams help

Each free question and every premium exam mirrors the scenario style Microsoft uses: a short stem with two or three constraints and several plausible Azure options. Detailed explanations cover not just why the right answer is right but why each distractor fails on exactly one stated constraint, so you learn the trade-offs rather than memorizing answers.

How to prepare for the AI-200 exam

A successful AI-200 plan pairs structured study with daily hands-on time writing Python against real Azure back-end services. A recommended approach:

1. Study the skills measured (2 to 3 weeks). Walk the official AI-200 study guide and the Microsoft Learn training modules it links for each skill area. Give the data-services domain the most room — at 25 to 30% it is the largest single block — then spread the rest evenly across the three 20 to 25% areas.
2. Containers and hosting labs (2 to 3 weeks). Build an image, push it to Azure Container Registry, and redeploy it by digest rather than by tag. Wire an ACR Task, run the same image on App Service with environment variables and Key Vault-backed secrets, then on Container Apps with revisions and a KEDA queue scaler, then on AKS from a manifest. Break each one on purpose and read the logs, events, and connectivity path back to the cause.
3. Data and messaging labs (2 to 3 weeks). Query Azure Cosmos DB for NoSQL from the SDK, watch Request Units move as you change the indexing policy and the consistency level, store embeddings and run vector similarity search, and run a change feed processor. Do the same shape of work on Azure Database for PostgreSQL with pgvector and metadata-filtered retrieval, and in Azure Managed Redis with expiry, invalidation, and a vector index. Then hand work between components with Service Bus queues, topics, and dead-letter handling, Event Grid filters and retries, and Azure Functions triggers and bindings.
4. Security, tracing, and troubleshooting (1 week). Rotate a secret in Key Vault and prove the app picks it up without a redeploy, serve settings from App Configuration with labels, instrument a service with the OpenTelemetry SDKs, and practise KQL until you can find one failed request across two services from memory. This domain decides a lot of otherwise-tied questions.
5. Practice exams (1 to 2 weeks). Take timed practice tests to surface weak areas. Detailed explanations on every option help you learn the reasoning rather than memorize answers. Aim for consistent scores above 80% before you schedule.

Recommended timeline

Plan on 6 to 10 weeks of focused study (8 to 12 hours per week) for developers who already have some Python and Azure experience. If you come from AZ-204, you can move quickly through containers, Functions, Key Vault, and monitoring, and should spend the extra time on the vector-search and data-store material: Cosmos DB vector search and change feed, pgvector indexing and RAG with metadata filters, and Redis vector indexing.

Official resources

Read the official AI-200 study guide for the exact skills measured, which were revised on April 15, 2026, and review the Azure AI Cloud Developer Associate certification page for registration and renewal details. Microsoft also publishes the AI-200T00 instructor-led course. A Microsoft practice assessment is not yet published for this exam — as of July 2026 the study guide's Useful links table has no practice-assessment row — and these usually appear within about eight weeks of an exam leaving beta, so lean on hands-on labs and these practice tests in the meantime.

Frequently asked questions