Azure-AI_Foundry

Lesson 2: Getting Started with Azure AI Foundry

Learning Objectives

By the end of this lesson, you will be able to:

Create an AI Foundry hub using Portal, CLI, Bicep, and SDK
Create a project within a hub
Navigate the AI Foundry portal
Set up local development with VS Code and the SDK

📖 Official docs: Create an AI Foundry project

Creating Your First Hub

Option 1: Azure Portal

Navigate to ai.azure.com
Click + Create project
Select New hub → configure name, region, resource group
Review dependent resources (Storage, Key Vault, AI Services)
Click Create

Option 2: Azure CLI

# Install the ML extension (AI Foundry uses the ML control plane)
az extension add --name ml --upgrade

# Create a resource group
az group create --name rg-ai-foundry-dev --location eastus

# Create an AI Hub
az ml workspace create \
  --kind hub \
  --name hub-myorg-dev \
  --resource-group rg-ai-foundry-dev \
  --location eastus

Option 3: Bicep (Infrastructure as Code)

// main.bicep — Create an AI Foundry Hub
@description('Name of the AI Hub')
param hubName string = 'hub-myorg-dev'

@description('Azure region for deployment')
param location string = resourceGroup().location

resource storageAccount 'Microsoft.Storage/storageAccounts@2023-05-01' = {
  name: 'st${uniqueString(resourceGroup().id)}'
  location: location
  sku: { name: 'Standard_LRS' }
  kind: 'StorageV2'
}

resource keyVault 'Microsoft.KeyVault/vaults@2023-07-01' = {
  name: 'kv-${uniqueString(resourceGroup().id)}'
  location: location
  properties: {
    tenantId: subscription().tenantId
    sku: { family: 'A', name: 'standard' }
    enableRbacAuthorization: true
  }
}

resource aiHub 'Microsoft.MachineLearningServices/workspaces@2024-04-01' = {
  name: hubName
  location: location
  kind: 'Hub'
  identity: { type: 'SystemAssigned' }
  properties: {
    friendlyName: hubName
    storageAccount: storageAccount.id
    keyVault: keyVault.id
  }
}

output hubId string = aiHub.id

# Deploy the Bicep template
az deployment group create \
  --resource-group rg-ai-foundry-dev \
  --template-file main.bicep

Option 4: Python SDK

from azure.ai.ml import MLClient
from azure.ai.ml.entities import Hub
from azure.identity import DefaultAzureCredential

ml_client = MLClient(
    credential=DefaultAzureCredential(),
    subscription_id="<subscription-id>",
    resource_group_name="rg-ai-foundry-dev"
)

hub = Hub(
    name="hub-myorg-dev",
    location="eastus",
    display_name="My Organization Dev Hub"
)

created_hub = ml_client.workspaces.begin_create(hub).result()
print(f"Hub created: {created_hub.name}")

Creating a Project Within a Hub

Azure CLI

az ml workspace create \
  --kind project \
  --name proj-chatbot \
  --resource-group rg-ai-foundry-dev \
  --hub-id /subscriptions/<sub>/resourceGroups/rg-ai-foundry-dev/providers/Microsoft.MachineLearningServices/workspaces/hub-myorg-dev

Python SDK

from azure.ai.ml.entities import Project

project = Project(
    name="proj-chatbot",
    hub_id=created_hub.id,
    display_name="Customer Chatbot"
)

created_project = ml_client.workspaces.begin_create(project).result()
print(f"Project created: {created_project.name}")

Navigating the AI Foundry Portal

The portal at ai.azure.com is organized into key sections:

Section	Purpose
Home	Recent projects, quick actions, learning resources
Model catalog	Browse and deploy 1,600+ models
Playgrounds	Interactively test models (Chat, Completions, Images, Audio)
Prompt flow	Build and test multi-step AI workflows
Evaluation	Run quality and safety evaluations
Deployments	Manage model endpoints
Data	Upload and manage datasets
Agents	Build AI agents with tools and data access

Portal Workflow

Model Catalog → Playground → Prompt Flow → Evaluation → Deployment
     ↓              ↓            ↓              ↓            ↓
  Select model   Prototype   Build pipeline   Validate    Go live

Setting Up VS Code Integration

1. Install Required Extensions

Azure Account — Sign in to Azure
Azure AI Foundry — Project management and debugging
Python — Python development support
Jupyter — Notebook support for experimentation

2. Install the SDK Packages

# Core packages for AI Foundry development
pip install azure-ai-projects          # Project management
pip install azure-ai-inference         # Model inference (chat, embeddings)
pip install azure-identity             # Authentication
pip install azure-ai-evaluation        # Evaluation framework

# Optional but recommended
pip install azure-ai-ml                # ML workspace management
pip install python-dotenv              # Environment variable management

3. Configure Authentication

# .env file — store your configuration
AZURE_SUBSCRIPTION_ID=your-subscription-id
AZURE_RESOURCE_GROUP=rg-ai-foundry-dev
AZURE_AI_HUB_NAME=hub-myorg-dev
AZURE_AI_PROJECT_NAME=proj-chatbot

# connect.py — verify your connection
import os
from dotenv import load_dotenv
from azure.identity import DefaultAzureCredential
from azure.ai.projects import AIProjectClient

load_dotenv()

project = AIProjectClient(
    credential=DefaultAzureCredential(),
    endpoint="https://hub-myorg-dev.services.ai.azure.com/api",
    project_name=os.getenv("AZURE_AI_PROJECT_NAME")
)

print("✅ Connected to AI Foundry project!")
print(f"   Project: {project._config.project_name}")

Understanding the SDK Ecosystem

Package	Purpose	Install
`azure-ai-projects`	Manage projects, connections, agents	`pip install azure-ai-projects`
`azure-ai-inference`	Call models (chat, embeddings, images)	`pip install azure-ai-inference`
`azure-ai-evaluation`	Evaluate model quality and safety	`pip install azure-ai-evaluation`
`azure-ai-ml`	Manage hubs, projects, compute (control plane)	`pip install azure-ai-ml`
`azure-identity`	Authentication (DefaultAzureCredential)	`pip install azure-identity`
`openai`	OpenAI-compatible API (works with Azure)	`pip install openai`

💡 Tips: Getting Started Best Practices

Use managed identity from day one — Avoid API keys; use DefaultAzureCredential for all SDK calls
Tag your resources — Apply environment, team, and project tags for cost and access tracking
Start in the playground — Prototype prompts interactively before writing code
Use .env files — Never hard-code connection strings or keys in source code
Enable diagnostic logging — Turn on Application Insights during hub creation
Version control your Bicep — Infrastructure as Code from the start

# ✅ Do this — managed identity
from azure.identity import DefaultAzureCredential
credential = DefaultAzureCredential()

# ❌ Not this — hardcoded keys
api_key = "sk-abc123..."  # Never do this!

Summary

Method	Best For
Portal	Quick experimentation and learning
CLI	Scripted setup and CI/CD pipelines
Bicep	Repeatable infrastructure deployments
SDK	Application-level resource management

Additional Resources

*← Lesson 1

Knowledge Check →*