AI-200 Develop AI Cloud Solutions on Microsoft Azure

This learning path guides you through core container hosting workflows on Azure for AI applications and backend services. You use Azure Container Registry to store and organize images, build images in the cloud with ACR Tasks, and apply tagging and versioning practices that support reliable deployments. From there, you deploy custom containers to Azure App Service, configure runtime behavior such as ports, startup commands, and persistent storage, and externalize environment-specific configuration using application settings.

Lesson:
• Store and manage containers in Azure Container Registry
o Use Azure Container Registry to store, build, and manage container images for AI applications. Learn the registry hierarchy, build images with ACR Tasks, and implement tagging strategies for reliable deployments.
• Deploy containers to Azure App Service
o Deploy and manage containerized applications on Azure App Service by configuring container sources, runtime settings, application configuration, and diagnostics.

This learning path guides you through the complete lifecycle of running containerized applications on Azure Container Apps. You start by deploying container apps to environments, configuring runtime settings with environment variables and secrets, and setting up registry authentication. You then manage the day-two lifecycle by updating images, managing revisions, monitoring logs, and configuring health probes. Finally, you learn to configure automatic horizontal scaling using HTTP rules, KEDA scalers, and traffic management to optimize performance and cost.

Lesson:
• Deploy containers to Azure Container Apps
o Create and manage container deployments in Azure Container Apps. Work with environments, runtime configuration, registry authentication, and deployment verification techniques.
• Manage containers in Azure Container Apps
o Manage container apps across the day-two lifecycle. Update images, manage revisions, diagnose failing deployments, tune resources and scaling, and troubleshoot with logs and health probes.
• Scale containers in Azure Container Apps
o Learn how to configure automatic horizontal scaling for containerized applications in Azure Container Apps. Configure HTTP, TCP, CPU, and memory scale rules, implement event-driven scaling with KEDA, optimize compute resources, and apply revision modes for traffic management.

This learning path guides you through the complete lifecycle of running applications on Azure Kubernetes Service. You start by creating deployment manifests and exposing applications with Kubernetes Services. You then externalize configuration using ConfigMaps, secure sensitive settings with Secrets, and attach persistent storage for stateful workloads. Finally, you learn to monitor application health using logs and metrics, troubleshoot pod and Service issues, and verify connectivity paths to ensure reliable access to your applications.

Lesson:
• Deploy applications to Azure Kubernetes Service
o Learn how to deploy applications to Azure Kubernetes Service. This module covers creating deployment manifests, exposing applications with services, and deploying to Azure Kubernetes Service.
• Configure applications on Azure Kubernetes Service
o Learn to externalize configuration, secure sensitive settings, and attach persistent storage using Kubernetes features on Azure Kubernetes Service.
• Monitor and troubleshoot applications on Azure Kubernetes Service
o Learn to monitor application health, inspect logs and metrics, troubleshoot pods and Services, and verify connectivity for AI workloads on Azure Kubernetes Service (AKS).

This learning path guides you through developing AI solutions using Azure Cosmos DB for NoSQL. You start by building a data foundation with the Cosmos DB resource model, SDK integration, CRUD operations, and SQL queries to retrieve document data for AI applications.

You then implement vector search capabilities to store embeddings, execute similarity queries using the VectorDistance function, combine vector search with metadata filters and hybrid search, and use the change feed to keep embeddings synchronized.

Finally, you optimize query performance by analyzing query patterns, configuring range and composite indexes, selecting vector index types, and choosing consistency levels that balance freshness with cost efficiency.

Lesson:
• Build queries for Azure Cosmos DB for NoSQL
o Learn how to connect to Azure Cosmos DB for NoSQL using the SDK, perform data operations on items, and write efficient SQL queries to retrieve document data for AI applications.
• Implement vector search on Azure Cosmos DB for NoSQL
o Learn how to store vector embeddings, execute similarity queries using the VectorDistance function, combine vector search with metadata filters and hybrid search, and use the change feed to keep embeddings synchronized
• Optimize query performance for Azure Cosmos DB for NoSQL
o Learn how to optimize query performance by analyzing query patterns, configuring range and composite indexes, selecting vector index types, and choosing consistency levels that balance freshness with cost efficiency.

This learning path guides you through developing AI solutions using Azure Database for PostgreSQL. You start by building a data foundation with schema design, efficient SQL queries, and secure Python integration using Microsoft Entra authentication.

You then implement vector search using the pgvector extension to store embeddings, execute similarity searches with different distance metrics, and build retrieval patterns that integrate with RAG pipelines for semantic search and recommendations.

Finally, you optimize vector search performance by tuning PostgreSQL and pgvector configuration, selecting appropriate vector indexes, designing efficient data layouts, scaling for high-volume workloads, and implementing connection pooling for AI applications.

Lesson:
• Build and query with Azure Database for PostgreSQL
o Learn how to use Azure Database for PostgreSQL to build data foundations for AI applications. Design schemas, write efficient queries, and integrate with Python applications using secure authentication.
• Implement vector search with Azure Database for PostgreSQL
o Learn how to implement vector search using the pgvector extension in Azure Database for PostgreSQL. Store embeddings, create vector indexes, and build semantic retrieval patterns for AI applications.
• Optimize vector search in Azure Database for PostgreSQL
o Learn how to optimize vector search performance in Azure Database for PostgreSQL using pgvector. Tune configuration parameters, select and configure vector indexes, design efficient data layouts, scale for high-volume workloads, and implement connection pooling for AI applications.

Learn how to use Azure Managed Redis to enhance your AI solutions, including caching strategies, data operations, event messaging, and vector storage.

Lesson:
• Implement data operations in Azure Managed Redis
o Learn how to implement data operations in Azure Managed Redis. This module covers Azure Managed Redis features, client library best practices, and how to store and retrieve data efficiently.
• Implement event messaging with Azure Managed Redis
o Learn how to implement event messaging with Azure Managed Redis, including pub/sub for broadcasting notifications and Redis Streams for reliable async task processing. This module covers building real-time notification systems and coordinating multi-step processing pipelines.
• Implement vector storage in Azure Managed Redis
o Learn how to implement vector storage and similarity search in Azure Managed Redis. This module covers creating vector indexes, querying embeddings, choosing vector types and indexing strategies, and selecting optimal data structures for AI applications.

This learning path teaches you how to build and integrate backend services that support AI solutions on Azure. You start by using Azure Service Bus to decouple AI application components, queue inference requests, and process messages reliably with queues, topics, and dead-letter queues. You then build event-driven workflows with Azure Event Grid to route events between services with low latency, configure delivery policies, and publish custom events from AI applications. Finally, you create serverless AI backends with Azure Functions, including inference endpoints, event processors, and secure integrations with other Azure services.

Lesson:
• Queue and process AI operations with Azure Service Bus
o Learn how to use Azure Service Bus to decouple AI application components, queue inference requests, distribute processing workloads across competing consumers, and handle failures through dead-letter queues. This module covers queues, topics with subscriptions, message structuring for AI payloads, and reliable message processing with the Python SDK.
• Develop event-driven AI workflows with Azure Event Grid
o Build reactive AI architectures using Azure Event Grid to route events from sources to handlers with low latency and high reliability. Learn to configure event subscriptions, design CloudEvents, implement delivery policies, and publish custom events from AI applications.
• Build serverless AI backends with Azure Functions
o Learn how to use Azure Functions as lightweight serverless compute for AI workloads. Build inference endpoints, event processors, and service integrations that scale automatically with demand.

This learning path teaches you how to securely manage secrets and centralize configuration for AI solutions on Azure. You start by using Azure Key Vault to store, organize, and retrieve secrets with managed identity authentication, handle secret versioning and rotation for zero-downtime credential updates, and implement caching strategies that reduce API calls while maintaining credential freshness. You then use Azure App Configuration to centralize application settings, organize key-value pairs with labels for environment-specific variants, manage feature flags for controlled rollouts, and reference Key Vault secrets so the application retrieves configuration and secrets through a single path.

Lesson:
• Manage application secrets with Azure Key Vault
o Learn how to use Azure Key Vault to store, retrieve, and manage secrets in AI solutions on Azure. This module covers vault organization, SDK-based secret retrieval with managed identity, secret versioning and rotation strategies, and caching patterns that reduce API calls while maintaining credential freshness.
• Manage application settings with Azure App Configuration
o Learn how to use Azure App Configuration to centralize application settings for AI solutions on Azure. This module covers connecting from Python application code with managed identity, organizing settings with labels and feature flags, referencing Key Vault secrets, and deciding which settings belong in each service.

This learning path teaches you how to gain end-to-end observability into distributed AI applications on Azure. You start by instrumenting applications with OpenTelemetry to capture distributed traces, create custom spans, and export telemetry to Azure Monitor Application Insights. You then analyze the collected telemetry by writing KQL queries, exploring error patterns and performance trends, building dashboards and workbooks for operational visibility, and configuring alerts to detect failures and anomalies.

Lesson:
• Instrument an app with OpenTelemetry
o Learn how to instrument distributed applications with OpenTelemetry on Azure, create custom spans and traces, export telemetry to Azure Monitor Application Insights, and use trace data to debug performance issues in distributed AI solutions.
• Analyze app telemetry with logs and metrics
o Learn to write KQL queries against Application Insights logs, explore error patterns and performance trends, build dashboards and workbooks for ongoing visibility, and configure alerts to detect failures and anomalies in AI solutions on Azure.