Azure OpenAI Workshop

A comprehensive hands-on workshop for building AI applications with Azure AI Foundry. Learn to deploy models, implement tracing & observability, and create intelligent agent-based applications.

What You'll Learn

• Deploy and test Azure OpenAI models
• Implement monitoring and tracing for AI applications
• Build intelligent agent-based applications
• Work with Azure AI Foundry SDK

Prerequisites

Before starting, ensure you have:

• Azure Account - Get free Azure credits
• Azure CLI - For authentication and infrastructure deployment
• Terraform - For infrastructure as code deployment
• Python 3.12+ - Required for the workshop environment
• UV modern Python dependency manager

Authenticate with Azure

# Login to Azure
az login

# Verify your subscription
az account show

Getting Started

Step 1: Clone and Setup Environment

Clone the repository:

git clone https://github.com/yanivvak/azure-openai-workshop.git
cd azure-openai-workshop

Install dependencies:

uv sync

Validate installation:

uv run python validate_setup.py

Step 2: Deploy Azure Infrastructure

Using Terraform (Multi-cloud):

Platform	Command
macOS/Linux	cd infrastructure/terraform && ./deploy.sh
Windows (PowerShell)	cd infrastructure/terraform; terraform init; terraform apply
Windows (Git Bash)	cd infrastructure/terraform && ./deploy.sh

This deployment creates:

• Azure AI Foundry resource
• AI Foundry project
• GPT-4.1-mini model deployment
• Required permissions and security settings

Step 3: Configure Environment

1. Copy environment template:

   Platform	Command
   macOS/Linux/Git Bash	cd ../../ && cp .env.example .env
   Windows (PowerShell)	cd ..\..; Copy-Item .env.example .env
   Windows (Command Prompt)	cd ..\.. && copy .env.example .env

2. Update required variables in .env using deployment outputs:

   # From deployment output: aiFoundryEndpoint
   AZURE_OPENAI_ENDPOINT=https://your-foundry-resource.cognitiveservices.azure.com/
   
   # From deployment output: modelDeploymentName  
   AZURE_OPENAI_DEPLOYMENT_NAME=gpt-4.1-mini
   
   # API version
   AZURE_OPENAI_API_VERSION=2024-10-21

3. Choose authentication method:

   Option A: Entra ID (Recommended)

   • Uses your az login credentials
   • No additional setup required

   Option B: API Key

   • Get key from Azure Portal → AI Foundry resource → Keys and Endpoint
   • Add to .env: AZURE_OPENAI_API_KEY=your-api-key

Workshop Notebooks (Core Workshop)

Notebook	Topic	Description
01-deploy-first-model.ipynb	Model Deployment	Deploy and test your first Azure OpenAI model
02-foundry-tracing.ipynb	Tracing & Observability	Implement monitoring and tracing for AI applications
03-agent_evaluators.ipynb	Agent Evaluation	Learn to evaluate and test AI agents
04-agents.ipynb	Agent Applications	Build intelligent agent-based applications
05-agents_tracing.ipynb	Advanced Tracing	Advanced tracing techniques for agents
06-fine_tuning.ipynb	Fine-tuning	Custom model fine-tuning with Azure OpenAI

Azure Data Explorer (ADX) Integration

For enterprise-scale security analytics and advanced monitoring, explore the ADX integration:

• Location: adx/ folder
• Notebook: adx/notebooks/03-security-pentesting-adx.ipynb
• Documentation: ADX README
• Features: Real-time analytics, cost tracking, security monitoring, and business intelligence

Troubleshooting

Validate Your Setup

uv run python validate_setup.py

This checks all required packages, Azure SDK integration, and environment configuration.

Common Issues

Issue	Solution
Authentication errors	Run az login and verify with az account show
Azure CLI not found	Install Azure CLI, then run az login
Wrong subscription	Run az account set --subscription "Your Subscription Name"
Module import errors	Run uv sync to reinstall dependencies
Environment variable errors	Ensure required variables are set in .env file
API quota exceeded	Request quota increase or change Azure region
Model deployment fails	Ensure S0 SKU and region supports GPT-4.1-mini

Quick Environment Check

Verify your environment variables:

macOS/Linux/Git Bash:

echo "Endpoint: $AZURE_OPENAI_ENDPOINT"
echo "Deployment: $AZURE_OPENAI_DEPLOYMENT_NAME" 
echo "API Version: $AZURE_OPENAI_API_VERSION"

Windows (PowerShell):

Write-Host "Endpoint: $env:AZURE_OPENAI_ENDPOINT"
Write-Host "Deployment: $env:AZURE_OPENAI_DEPLOYMENT_NAME"
Write-Host "API Version: $env:AZURE_OPENAI_API_VERSION"

Windows (Command Prompt):

echo Endpoint: %AZURE_OPENAI_ENDPOINT%
echo Deployment: %AZURE_OPENAI_DEPLOYMENT_NAME%
echo API Version: %AZURE_OPENAI_API_VERSION%

Advanced Configuration

Adding Dependencies

# Add a new package
uv add package-name

# Add development tools  
uv add --group dev package-name

Included Dependencies

The workshop includes all necessary packages:

• Azure AI: azure-ai-projects[agents], azure-ai-inference, azure-identity
• OpenAI SDK: openai with Azure OpenAI support
• Observability: opentelemetry-*, azure-monitor-opentelemetry
• Data Analytics: azure-kusto-data, azure-kusto-ingest (Azure Data Explorer)
• Data & Analysis: pandas, numpy, matplotlib
• Jupyter: jupyter, ipykernel

Additional Resources

• Azure AI Foundry Documentation
• Infrastructure Setup Guide
• Azure OpenAI Service Documentation

Contributing

Contributions are welcome! Please feel free to submit issues and enhancement requests.

License

This project is licensed under the MIT License - see the LICENSE file for details.