Architecture (Python Inference Service)

Technology Stack

• Framework: FastAPI 0.110.0 (async Python web framework)
• Server: Uvicorn 0.28.0 (ASGI server)
• ML Framework: PyTorch 2.2.1
• Numerical Computing: NumPy 1.26.4
• Validation: Pydantic 2.6.4
• Templating: Jinja2 3.1.3
• Environment: Python 3.9

Deployment

• Container: Docker (Python 3.9-slim base)
• Port: 8000 (configurable via API_PORT)
• Production URL: https://traffic-inference-service.onrender.com
• Local Development: http://localhost:8000