Training: Develop AI cloud solutions on Azure (AI-200)

Ref. AI-200T00

Download PDF

Duration:

jours

Exam:

Optionnel

Level:

Intermédiaire

Funding:

Eligible

Home > Trainings > Artificial intelligence > AI Development and LLM > Azure AI Platforms > Develop AI cloud solutions on Azure (AI-200)

Develop AI cloud solutions on Azure Training (AI-200)

Développer des solutions IA cloud robustes dans Azure exige de maîtriser à la fois les services Azure AI, l’architecture cloud-native et les bonnes pratiques de mise en production. La formation AI-200 couvre l’ensemble du cycle de vie : conception, développement, déploiement et exploitation de solutions IA génératives et conventionnelles dans Azure.

Pendant cinq jours, vous travaillez sur Azure AI Foundry, Azure OpenAI Service, Azure AI Search, les agents intelligents, le RAG, et l’intégration avec les services data Azure (Cosmos DB, SQL, Storage). La formation est animée à Genève et Lausanne par un formateur Microsoft Certified Trainer.

Participant Profiles

Objectives

Architect an end-to-end AI solution in Azure following Well-Architected Framework principles
Develop generative applications with Azure OpenAI Service and the Azure AI SDK
Build AI agents orchestrating tools, data sources and workflows
Implement complete RAG pipelines with Azure AI Search and vector embeddings
Industrialize AI deployment with Azure MLOps (model registry, monitoring, drift detection)
Secure cloud AI solutions with content safety, managed identity and private networking

Prerequisites

Course Content

Module 1 : Store and manage containers in Azure Container Registry

Registries, repositories, and artifacts
Build and run images with ACR Tasks
Tag and version images
Module assessment

Module 2 : Deploy containers to Azure App Service

Deploy containers to Azure App Service
Configure container runtime behavior
Configure application settings
Observe and troubleshoot containerized apps
Module assessment

Module 3 : Deploy containers to Azure Container Apps

Explore Container Apps environments
Deploy a container app using the Azure CLI and YAML
Configure runtime settings with environment variables and secrets
Configure image pull authentication for private registries
Verify deployments with logs and status
Module assessment

Module 4 : Manage containers in Azure Container Apps

Update images and manage revisions safely
Manage the container app lifecycle
Monitor logs and troubleshoot issues
Configure health probes and troubleshoot failures
Optimize container resources and scaling
Module assessment

Module 5 : Scale containers in Azure Container Apps

Configure scale rules
Implement event-driven scaling with KEDA
Apply KEDA scalers for custom workloads
Select compute resources for performance and cost
Choose and apply revision modes
Module assessment

Module 6 : Deploy applications to Azure Kubernetes Service

Create Kubernetes deployment manifests
Expose applications in Azure Kubernetes Services
Deploy applications to Azure Kubernetes Services
Module assessment

Module 7 : Configure applications on Azure Kubernetes Service

Define ConfigMaps for application settings
Implement secrets for sensitive data
Attach persistent storage to an app
Module assessment

Module 8 : Monitor and troubleshoot applications on Azure Kubernetes Service

Monitor application logs and metrics
Troubleshoot pods and services
Verify service connectivity and endpoints
Module assessment

Module 9 : Build queries for Azure Cosmos DB for NoSQL

Explore Azure Cosmos DB for NoSQL
Implement the Azure Cosmos DB for NoSQL SDK
Query Azure Cosmos DB for NoSQL
Module assessment

Module 10 : Implement vector search on Azure Cosmos DB for NoSQL

Store and retrieve embeddings in Azure Cosmos DB
Execute vector similarity queries for semantic search
Combine vector similarity results with metadata filtering
Use the change feed to trigger embedding refresh
Module assessment

Module 11 : Optimize query performance for Azure Cosmos DB for NoSQL

Understand indexes in Azure Cosmos DB
Configure range and composite indexes
Tune vector indexes for embedding workloads
Reduce RU costs with strategic indexing
Choose consistency levels for optimal performance
Module assessment

Module 12 : Build and query with Azure Database for PostgreSQL

Explore Azure Database for PostgreSQL
Connect to PostgreSQL
Create and manage schemas
Query data
Integrate SDKs and applications
Module assessment

Module 13 : Implement vector search with Azure Database for PostgreSQL

Store and query embeddings with pgvector
Perform fast vector similarity search
Manage index lifecycle and embedding updates
Run vector similarity search for semantic retrieval
Implement retrieval patterns for RAG pipelines
Module assessment

Module 14 : Optimize vector search in Azure Database for PostgreSQL

Tune PostgreSQL for pgvector
Choose and configure vector indexes
Optimize data layout
Scale for high-volume workloads
Connection optimization
Module assessment

Module 15 : Implement data operations in Azure Managed Redis

Explore Azure Managed Redis
Client libraries and development best practices
Implement data operations
Module assessment

Module 16 : Implement event messaging with Azure Managed Redis

Publish and subscribe to events with Redis pub/sub
Implement task queues with Redis Streams
Choose between broadcast and coordinated distribution
Module assessment

Module 17 : Implement vector storage in Azure Managed Redis

Index and query vector data
Choose vector types and indexing strategies
Optimize Redis data structures for vector storage
Module assessment

Module 18 : Queue and process AI operations with Azure Service Bus

Explore Azure Service Bus concepts and messaging in AI architectures
Choose between queues and topics with subscriptions
Structure messages for AI workloads
Process messages reliably
Module assessment

Module 19 : Develop event-driven AI workflows with Azure Event Grid

Understand Azure Event Grid concepts and event-driven patterns for AI solutions
Work with event schemas and properties
Configure delivery and retry policies for reliable event processing
Publish custom events from AI applications
Module assessment

Module 20 : Build serverless AI backends with Azure Functions

Understand Azure Functions hosting and scaling for AI workloads
Set up the local development environment for Functions
Create triggers and bindings for AI integration patterns
Manage secrets and configuration in Functions
Configure identity and access for Functions
Module assessment

Module 21 : Manage application secrets with Azure Key Vault

Store and organize secrets, keys, and certificates
Retrieve secrets using Azure SDK client libraries
Handle secret versioning and rotation
Implement caching strategies to reduce Key Vault calls
Module assessment

Module 22 : Manage application settings with Azure App Configuration

Connect to App Configuration from application code
Organize settings with labels and feature flags
Reference Key Vault secrets from App Configuration
Decide what to store in App Configuration vs Key Vault
Module assessment

Module 23 : Instrument an app with OpenTelemetry

Explore OpenTelemetry and its role in observability
Add the OpenTelemetry SDK to an application
Configure spans and traces
Export telemetry to Azure Monitor
Debug distributed flows with trace data
Module assessment

Module 24 : Analyze app telemetry with logs and metrics

Write basic KQL queries
Explore logs for errors and performance
Build dashboards for app telemetry
Create workbooks for interactive analysis
Set alerts for app failures and anomalies
Module assessment

Documentation

Course material included.

Exam

This course prepares you to the AI-102: Designing and Implementing a Microsoft Azure AI Solution exam.

Complementary Courses

Eligible Funding

ITTA is a partner of a continuing education fund dedicated to temporary workers. This fund can subsidize your training, provided that you are subject to the “Service Provision” collective labor agreement (CCT) and meet certain conditions, including having worked at least 88 hours in the past 12 months.

Additional Information

Why the Develop AI cloud solutions on Azure (AI-200) training

Companies are industrializing their AI initiatives and demand architects able to design robust cloud-native AI solutions. AI-200 goes beyond Jupyter experimentation: you address production, observability, security and compliance. The course is for those who transform AI proof-of-concepts into scalable services in production on Azure.

Azure AI Foundry and foundation models

Azure AI Foundry is the development hub. The training has you deploy models from the model catalog (GPT-4o, Phi, Mistral, Llama), evaluate their performance on your datasets, and orchestrate complex workflows with prompt flow. You also work on supervised fine-tuning to adapt models to your business use cases.

AI agents and orchestrated architectures

The program covers agent design with Azure AI Agent Service and multi-agent patterns. You build agents able to reason on plans, invoke function calls to external APIs, and cooperate to solve complex tasks. Design principles (ReAct, Plan-and-Execute, tool-augmented LLM) are detailed.

Enterprise-grade RAG with Azure AI Search

RAG is treated in depth: hybrid vector + lexical search, semantic ranking, query rewriting, contextual chunking, and systematic evaluation with groundedness and relevance metrics. You compare strategies (basic RAG vs agentic RAG vs graph RAG) and choose according to the use case.

MLOps and industrialization

AI industrialization is not optional: the training addresses Azure Machine Learning model registry, production monitoring (latency, cost, drift), automated retraining and CI/CD integration with Azure DevOps and GitHub Actions. The goal is to reach a reproducible and auditable AI pipeline.

Security, identity and private networking

Securing a cloud AI solution goes through Managed Identities, private endpoints (to avoid public traffic), Azure Key Vault for secrets, and Content Safety to filter LLM outputs. The training also covers conditional access controls and Microsoft Entra ID integration.

Audience and prerequisites

The Develop AI cloud solutions on Azure (AI-200) training targets solution architects, experienced AI developers and ML engineers who will design production AI systems. Prerequisites: Python experience, Azure fundamentals (AZ-900), AI knowledge (AI-900). Prior experience with Azure OpenAI or Azure AI Search is a plus.

FAQ Develop AI cloud solutions on Azure (AI-200)

What’s the difference between AI-200 and AI-102 / AI-103?

AI-200 is more architecture and end-to-end solution oriented (5 days), while AI-103 / AI-102 are oriented toward application development (4 days). AI-200 includes MLOps, advanced security and enterprise multi-agent architectures.

Does the training cover Azure Machine Learning?

Yes, with MLOps focus: model registry, managed online endpoints deployment, monitoring and CI/CD integration. Classic data science (model training) is addressed transversally.

Do I need prior AI experience to take AI-200?

Prior experience of at least 6 months in AI development or Azure data engineering is strongly recommended. AI-103 or AI-102 is an entry path.

Does the AI-200 course lead to a Microsoft certification?

AI-200 prepares for the Microsoft Certified: Azure AI Engineer Associate certification (AI-102 exam). The training goes beyond the exam program and covers architecture topics not formally tested.