Official curriculum guide

AI-200 Study Modules

9 comprehensive engineering modules mapping 100% of the official Microsoft Course AI-200T00-A: Develop AI Cloud Solutions on Azure curriculum. Each module provides production architecture breakdowns, verified Azure CLI & SDK snippets, failure point analyses, and exam traps.

Curriculum Architecture & Domain Overview Author: Rohan (Cloud & Infrastructure Software Engineer) · Technical Advisory Board Verified

The Microsoft Certified: Azure AI Cloud Developer Associate (AI-200) certification marks an intentional evolution from traditional generic application hosting (such as AZ-204) toward enterprise AI platform engineering. Modern cloud developers must master how to reliably host containerized microservices, build scalable vector retrieval backends, ingest high-frequency telemetry events, and protect distributed credentials across complex multi-service topologies. Rather than evaluating generic code syntax, the exam challenges candidates on real-world systems architecture, configuration trade-offs, and operational failure modes.

The curriculum is structured around four high-weight domains. Domain 1 (20–25%) focuses on containerized application hosting across Azure Container Registry (ACR), Azure Container Apps (ACA), and Azure Kubernetes Service (AKS). Candidates are evaluated on multi-revision blue-green cutovers, KEDA-driven event scaling, and federating cluster identities using Microsoft Entra Workload Identity. Domain 2 (30–35%) is the highest-weighted section of the exam, covering AI data management services. It demands deep technical proficiency with vector embeddings, similarity search algorithms, and indexing mechanics across Azure Cosmos DB for NoSQL (DiskANN vs. Flat), Azure Database for PostgreSQL Flexible Server (pgvector HNSW vs. IVFFlat), and Azure Managed Redis semantic caching.

Domain 3 (20–25%) tests asynchronous decoupling and serverless event pipelines using Azure Service Bus, Event Grid, and Azure Functions. Key exam scenarios center on FIFO session ordering, dead-letter queue (DLQ) diagnostic patterns, and reactive event filters. Finally, Domain 4 (20–25%) covers zero-trust cloud security and distributed observability. Candidates must demonstrate proficiency in automated secret rotation with Azure Key Vault, centralized feature management via Azure App Configuration, and distributed tracing with OpenTelemetry SDKs and Kusto Query Language (KQL) in Azure Monitor.

Official AI-200 Exam Domain Weights & Service Mapping (October 2026 Revision)

Domain Title Exam Weight Core Azure Services Primary Architectural Competencies
Domain 1 Develop Containerized Solutions on Azure 20–25% ACR, Container Apps, AKS, App Service OIDC Workload Identity, KEDA event scaling, ACR Tasks, canary revisions
Domain 2 Develop AI Solutions by Using Azure Data Management Services 30–35% (Highest) Cosmos DB NoSQL, PostgreSQL Flexible Server, Managed Redis DiskANN indexing, pgvector HNSW/IVFFlat, RediSearch semantic caching, Change Feed
Domain 3 Connect to and Consume Azure Services 20–25% Service Bus, Event Grid, Event Hubs, Azure Functions FIFO message sessions, DLQ triage, reactive push routing, serverless bindings
Domain 4 Secure, Monitor, and Troubleshoot Azure Solutions 20–25% Key Vault, App Configuration, Azure Monitor, Application Insights Azure RBAC, automated secret rotation, OpenTelemetry W3C tracing, KQL queries
Study Module Progress 0 / 9 completed
Domain 1

Develop Containerized Solutions on Azure

(20–25%)
Module 1 20–25%

Implement Container Application Hosting on Azure

Master container lifecycle management on Azure Container Registry. Configure Basic, Standard, and Premium SKUs with geo-replication for global AKS clusters. Automate cloud-based Docker builds, OS patching, and base-image updates with ACR Tasks using double-curly Go variables. Implement zero-trust security by replacing administrative credentials with Microsoft Entra Managed Identities and AcrPull role assignments. Connect containerized workloads to Azure App Service with secure environment variables and secrets.

Azure Container RegistryACR TasksAzure App Service
6 topics Study now →
Module 2 20–25%

Deploy and Manage Apps on Azure Container Apps

Deploy and manage microservices on serverless Azure Container Apps. Master single versus multiple revision modes for canary deployments, blue-green cutovers, and traffic-splitting experiments. Implement event-driven autoscaling using Kubernetes Event-driven Autoscaling (KEDA) scalers targeting Azure Service Bus queues and HTTP concurrency limits. Configure secure internal virtual networks, custom domains, managed certificate binding, and multi-container sidecar architectures.

Azure Container AppsKEDAAKS
5 topics Study now →
Module 3 20–25%

Deploy and Monitor Applications on Azure Kubernetes Service

Operate enterprise Kubernetes clusters on Azure Kubernetes Service. Configure system and user node pools, compute autoscaling, and network architectures using Azure CNI Overlay. Eliminate node-level credentials by implementing Microsoft Entra Workload Identity federated with Kubernetes Service Accounts via OIDC. Deploy production stateful and stateless workloads using declarative YAML manifests, and troubleshoot runtime pod failures using Container Insights.

Azure Kubernetes ServicekubectlHelm
5 topics Study now →
Domain 2

Develop AI Solutions by Using Azure Data Management Services

(30–35%)
Module 4 30–35%

Develop AI Solutions with Azure Cosmos DB for NoSQL

Architect high-performance distributed vector databases with Azure Cosmos DB for NoSQL. Configure container indexing policies using Flat and DiskANN approximate nearest neighbor (ANN) vector search. Tune partition keys and request units (RU/s) to avoid 429 rate-limiting during high-throughput semantic ingestion. Implement the Change Feed Processor pattern to trigger real-time AI embedding pipelines as documents are inserted or updated.

Azure Cosmos DBDiskANNChange Feed
7 topics Study now →
Module 5 30–35%

Develop AI Solutions with Azure Database for PostgreSQL

Deploy enterprise relational and vector retrieval workloads on Azure Database for PostgreSQL Flexible Server. Install and configure the pgvector extension, choosing between IVFFlat (faster build, lower memory) and HNSW (higher recall, zero retraining) indexing strategies. Execute cosine (<=>), inner product (<#>), and L2 distance (<->) vector similarity queries. Implement connection pooling with PgBouncer to eliminate connection exhaustion under load.

Azure Database for PostgreSQLpgvectorHNSW
6 topics Study now →
Module 6 30–35%

Enhance AI Solutions with Azure Managed Redis

Implement low-latency semantic caching and distributed state management using Azure Managed Redis. Build vector indexes using RediSearch to cache LLM inference outputs and slash foundation model API costs. Configure Redis active-active geo-replication across multi-region Azure deployments with Conflict-Free Replicated Data Types (CRDTs). Apply fine-grained expiration, TTL eviction policies, and cluster sharding strategies.

Azure Managed RedisRediSearchVector Index
4 topics Study now →
Domain 3

Connect to and Consume Azure Services

(20–25%)
Module 7 20–25%

Integrate Backend Services for AI Solutions

Build resilient event-driven architectures connecting decoupled microservices. Implement Azure Service Bus queues and topics with FIFO sessions, duplicate detection windows, dead-letter queue (DLQ) deadlocks, and SQL correlation filters. Route reactive lightweight events using Azure Event Grid system and custom topics. Build serverless backend APIs with Azure Functions using Cosmos DB, Service Bus, and Blob storage triggers.

Azure Service BusAzure Event GridAzure Functions
6 topics Study now →
Domain 4

Secure, Monitor, and Troubleshoot Azure Solutions

(20–25%)
Module 8 20–25%

Manage Application Secrets and Configuration for AI Solutions

Enforce zero-trust security across enterprise cloud solutions. Store and rotate cryptographic keys, certificates, and application secrets using Azure Key Vault with Azure RBAC governance. Manage dynamic feature toggles and centralized runtime settings via Azure App Configuration. Integrate Key Vault references into Container Apps and AKS pods using the Secrets Store CSI Driver.

Azure Key VaultAzure App Configuration
5 topics Study now →
Module 9 20–25%

Observe and Troubleshoot Apps on Azure

Instrument distributed cloud applications for full-stack observability. Integrate OpenTelemetry SDKs into Python and Node.js microservices with W3C TraceContext propagation across asynchronous message queues. Export distributed traces and metrics to Azure Monitor and Application Insights. Author advanced Kusto Query Language (KQL) forensics queries to diagnose cross-service latency bottlenecks and application exceptions.

OpenTelemetryKQLAzure Monitor
5 topics Study now →

Exam strategy & guides

Latest AI-200 Engineering Benchmarks & Articles

Architecture breakdowns, KQL cheatsheets, and real-world exam notes to accompany the modules.

Browse all 15 articles →
Exam Strategy September 2026

Strategic Guide: Using Open-Book Microsoft Learn on the AI-200 Exam

How to use Microsoft Learn during your 100-minute AI-200 exam without running out of time. What is fast to search versus what you must memorize.

Read guide →
Hands-On Lab September 2026

How to Run All AI-200 Hands-On Labs Locally for $0 (ACR Tasks Free Workaround)

Complete Docker Compose local testing harness for AI-200: emulate Cosmos DB, pgvector, Azure Managed Redis, and Azurite locally without spending Azure credits.

Read guide →
Exam Experience August 2026

I Failed AI-200 With 626/700. Here’s What I’d Do Differently

I took the AI-200 exam and did not pass by 74 points. Here is what the test was like, where I lost marks, and why I created ai200prep.com.

Read guide →
Hands-On Lab August 2026

Hands-On Lab: Zero-Trust AKS Workload Identity with Key Vault CSI Driver

Step-by-step production lab: setting up OIDC federated credentials, mounting secrets without pod restarts, and configuring Prometheus monitoring for AI-200.

Read guide →
Hands-On Lab August 2026

Hands-On Lab: Chunking Strategies & Vector Search Latency on pgvector vs DiskANN

Benchmarking 256 vs 512 vs 1024 token chunks with text-embedding-3-small across PostgreSQL pgvector and Cosmos DB DiskANN. Latency tables, QPS, and lab findings.

Read guide →
Technical Deep Dive July 2026

AKS vs Azure Container Apps: Choosing Correctly for AI-200

Scaling, networking, GPU support, cost model, and Dapr compared — plus the scenario phrases that determine which container platform is the expected answer.

Read guide →