Enterprise-ready solution for OpenAI usage tracking and chargeback through Azure APIM. Solves the 8KB payload limit problem, provides real-time cost tracking, and includes comprehensive Infrastructure-as-Code deployment options.
🚀 Quick Deploy: git clone → setup Python → deploy Bicep → configure policies
💰 What you get: Real-time usage tracking, chargeback reports, scalable architecture, multi-IaC support
🏗️ Tech Stack: Azure Functions (Python), API Management, Redis Cache, Bicep/Terraform
# 1. Clone and setup
git clone https://github.com/your-org/apim-openai-chargeback-environment.git
cd apim-openai-chargeback-environment
# 2. Setup Python environment
cd function-app && python -m venv venv && source venv/bin/activate
pip install -r requirements.txt
# 3. Deploy infrastructure (Bicep - recommended)
cd ../infrastructure/bicep
az deployment group create \
--resource-group myResourceGroup \
--template-file main.bicep \
--parameters @parameters/dev.json
# 4. Deploy Function App
cd ../../function-app && func azure functionapp publish <function-app-name>
# 5. Done! 🎉⚡ Alternative deployment options: Terraform | ARM Templates | Pulumi
| Challenge | Impact | Our Solution |
|---|---|---|
| APIM logs truncate at 8KB | Large OpenAI payloads get cut off | ✅ Handle payloads up to 100MB |
| No usage tracking | Cost overruns, no chargeback | ✅ Real-time cost tracking & reports |
| Manual deployment | Inconsistent environments | ✅ Multiple IaC options (Bicep, Terraform) |
| Security concerns | Data persistence issues | ✅ Minimal retention, encryption at rest |
graph LR
A[Client Apps] --> B[APIM Gateway]
B --> C[Azure Functions]
C --> D[Azure OpenAI]
C --> E[Redis Cache]
E --> F[Chargeback Reports]
Core Components:
- 🔐 API Management - Gateway, auth, rate limiting
- ⚡ Azure Functions (Python) - Usage tracking, cost calculation
- 💾 Redis Cache - Temporary data storage
- 📊 Chargeback Engine - Real-time cost allocation
📐 Need detailed architecture? See our comprehensive Architecture Guide with detailed component diagrams, data flows, and system design.
- 🚀 Scalable: Handles large payloads (>8192 bytes)
- 💰 Cost Tracking: Real-time usage monitoring and chargeback
- 🔒 Secure: Minimal data persistence, Azure AD integration
- 🏗️ Multi-IaC: Bicep (primary), Terraform, ARM templates, Pulumi
- 📈 Enterprise Ready: Multi-region, HA, comprehensive monitoring
- 🔄 CI/CD Ready: GitHub Actions, Azure DevOps pipelines
| Topic | Description | Link |
|---|---|---|
| Architecture | Detailed system design and components | 📖 Architecture Guide |
| Deployment | Step-by-step deployment for all IaC tools | 🚀 Deployment Guide |
| API Reference | Complete API documentation | 📚 API Docs |
| Usage Examples | Code examples in multiple languages | 💻 Usage Examples |
| Monitoring | Dashboards, alerts, and troubleshooting | 📊 Monitoring Guide |
| Security | Security best practices and compliance | 🔒 Security Guide |
| Contributing | Development setup and guidelines | 🤝 Contributing |
import requests
response = requests.post(
"https://your-apim.azure-api.net/api/v1/openai/chat/completions",
headers={"Ocp-Apim-Subscription-Key": "your-key"},
json={"model": "gpt-4", "messages": [{"role": "user", "content": "Hello!"}]}
)- C# / .NET: View Example
- Node.js: View Example
- PowerShell: View Example
- curl: View Example
Real-time usage tracking with detailed cost breakdown by user, model, and time period.
- ✅ Azure subscription with appropriate permissions
- ✅ Azure CLI v2.50+
- ✅ Python 3.9+
- ✅ Azure Functions Core Tools v4.x
- ✅ Docker Desktop (for local testing)
Optional: Bicep CLI, Terraform, or ARM templates depending on your IaC preference
# Essential configuration
AZURE_OPENAI_ENDPOINT=https://your-openai.openai.azure.com/
REDIS_CONNECTION_STRING=your-redis-connection-string
APIM_SUBSCRIPTION_KEY=your-apim-subscription-key📋 Complete configuration guide: Configuration Reference
├── infrastructure/ # IaC templates
│ ├── bicep/ # 🎯 Primary (recommended)
│ ├── terraform/ # Alternative
│ ├── arm/ # Alternative
│ └── pulumi/ # Alternative
├── function-app/ # Python Azure Functions
├── policies/ # APIM policies
├── scripts/ # Deployment scripts
└── docs/ # Detailed documentation
📝 Note: The app works as-is, but we're tidying up the structure. Work in progress! 🚧
| Method | Best For | Command |
|---|---|---|
| Bicep ⭐ | Azure-native teams | az deployment group create --template-file main.bicep |
| Terraform | Multi-cloud or existing Terraform | terraform apply -var-file="dev.tfvars" |
| ARM Templates | Legacy Azure deployments | az deployment group create --template-file azuredeploy.json |
| Pulumi | Programming-first approach | pulumi up |
🔗 Detailed deployment guides: All deployment methods
❓ What's the max payload size? ✅ Up to 100MB (vs 8KB APIM limit)
❓ How long is data retained?
✅ 30 days by default (configurable)
❓ Multi-region support? ✅ Yes, built for HA and disaster recovery
❓ Can I use .NET instead of Python? ✅ Yes, Azure Functions support multiple runtimes
📚 More questions: Complete FAQ
- 📊 Real-time dashboards for usage, costs, and performance
- 🚨 Automated alerts for anomalies and thresholds
- 📈 Custom metrics for business KPIs
- 🔍 Detailed logging for troubleshooting
🔧 Setup guide: Monitoring Configuration
- 📖 Documentation: Comprehensive guides in
/docs/ - 🐛 Issues: GitHub Issues
- 💬 Discussions: GitHub Discussions
- 🏢 Enterprise: Contact support team for professional services
| Quarter | Features |
|---|---|
| Q2 2025 | Multi-region deployment, advanced analytics |
| Q3 2025 | ML-based cost optimization, enhanced reporting |
| Q4 2025 | Additional AI service integrations |
We welcome contributions! See our Contributing Guide for:
- 🛠️ Development setup
- 🧪 Testing guidelines
- 📋 Code standards
- 🔄 PR process
MIT License - see LICENSE for details.
🎯 Ready to get started? Follow the Quick Start above or dive into the Deployment Guide.