Skip to content

Reference architecture that provides a set of guidelines and best practices for implementing a central AI API gateway to empower various line-of-business units in an organization to leverage Azure AI services

License

Notifications You must be signed in to change notification settings

sherskhan/ai-hub-gateway-solution-accelerator

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🚀 AI Hub Gateway Landing Zone

Enterprise-ready solution accelerator for implementing a centralized AI API gateway that empowers organizations to securely leverage multiple Azure AI services with unified governance, monitoring, and cost management.

AI Hub Gateway Landing Zone

⭐ What's New (Latest Updates)

🔒 Enterprise Security & Compliance

🧠 Expanded AI Service Portfolio

📊 Advanced Monitoring & Operations

🧩 Use Case Onboarding Automation

  • APIM Product + Subscription + KV Secrets (Bicep) - Automate per-use-case onboarding to the AI Gateway; creates per-service products, subscriptions, and writes endpoint + key secrets to Key Vault. Includes a ready-to-use Financial Assistant example.

🎯 Core Capabilities

ai-hub-gateway-benefits.png

🏢 Enterprise Governance

  • Centralized access control and API key management
  • Managed identity integration (no master keys required)
  • Multi-tenant isolation with product-based access control
  • Per-use-case onboarding automation for APIM Products and Subscriptions

⚡ Intelligent Routing

  • Priority-based backend selection with automatic failover
  • Regional load balancing across multiple AI backend instances
  • Capacity-aware routing with dynamic throttling for PTU models

💰 Cost Management

  • Real-time usage tracking and charge-back allocation
  • Token/Requests-level monitoring across all AI services
  • Flexible json based usage data model that supports extension
  • Power BI integration for self-service advanced analytics and reporting

🔐 Security & Compliance

  • Private endpoint connectivity for all managed services services
  • Network isolation with VNet integration
  • Enterprise authentication with Entra ID
  • PII detection and processing
  • LLM content safety for prompt and content filtering

one-click-deploy One-click Deploy

Deploy enterprise-ready AI governance in minutes with Azure Developer CLI (azd) or Bicep templates.

🏗️ What Gets Deployed

Azure components

Component Purpose Enterprise Features
🚪 API Management Central AI gateway with intelligent routing Load balancing, throttling, JWT validation
📊 Application Insights Real-time monitoring & analytics Custom dashboards, throttling alerts
📨 Event Hub Usage data streaming & processing Real-time cost tracking, compliance logging
🤖 Azure OpenAI Multi-region AI deployments (3 regions) GPT-models, Realtime API, fully private
🛡️ Azure Content Safety Centralized LLM protection Prompt Shield and Content Safety protections
💳 Azure Language Service PII entity detection Natural language based PII entity detection, anonymization
🗄️ Cosmos DB Usage analytics & cost allocation Global distribution, automatic scaling
⚡ Logic App Event processing & data transformation Workflow-based processing
🔐 Managed Identity Zero-credential authentication Secure service-to-service communication
🔗 Virtual Network Private connectivity & isolation BYOVNET support, private endpoints

📋 Prerequisites

Azure Requirements:

  • Azure Account with OpenAI access approved
  • Subscription with Microsoft.Authorization/roleAssignments/write permissions
  • Sufficient OpenAI capacity in target regions (East US, North Central US, East US 2)

Development Tools:

🚀 Quick Deploy

Review the main.bicep configuration, then deploy:

# Authenticate and setup environment
azd auth login
azd env new ai-hub-gateway-dev

# Deploy everything
azd up

💡 Tip: Use Azure Cloud Shell to avoid local setup. If deployment fails, retry azd up - it may be a transient error.

Once deployed, access your AI Gateway through the Azure API Management portal:

apim-test

docs Supporting Documents

Comprehensive guides to master AI Hub Gateway implementation and operations.

🏗️ Architecture & Deployment

Guide Description
Architecture Overview Complete system design and component relationships
Deployment Guide Step-by-step deployment instructions
Enterprise Provisioning NEW: Branch-based deployment strategy, parameter management, and CI/CD automation
APIM Configuration Advanced API Management policies and routing
Bring Your Own Network Deploy into existing VNets
Deployment Troubleshooting Common issues and solutions

🔧 Service Integration

Guide Description
OpenAI Onboarding Add new OpenAI instances and models
AI Search Integration Vector search and RAG capabilities
AI Foundry Integration Custom model deployment
End-to-end Scenario Complete chat-with-data implementation

🛡️ Security & Compliance

Guide Description
PII Detection & Masking Automated data protection
Entra ID Authentication JWT validation and Zero Trust
Use Case Onboarding Multi-service AI solution patterns

📊 Monitoring & Analytics

Guide Description
Power BI Dashboard Usage analytics and cost allocation
Throttling Events Real-time 429 error monitoring
Dynamic Throttling Intelligent load balancing
Usage Ingestion Token tracking and billing

⚙️ Advanced Features

Guide Description
Hybrid Deployment Multi-cloud and edge scenarios
Use Case Onboarding (APIM Product Automation) Automate per-use-case APIM Products, Subscriptions, and Key Vault secrets; includes “Financial Assistant” example

About

Reference architecture that provides a set of guidelines and best practices for implementing a central AI API gateway to empower various line-of-business units in an organization to leverage Azure AI services

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Bicep 95.8%
  • C# 1.8%
  • Shell 1.5%
  • PowerShell 0.9%