Model Jacket

Model Jacket is a lightweight framework that wraps any ML model with a FastAPI server, Docker image, Prometheus metrics, and Kubernetes Helm chart—so you can go from model file ➜ production endpoint in minutes.

Quick Start

pip install -e .
# Train or load a model and save as TorchScript
python export_model.py  # produces model.pt

jacket build --framework torch --model-path model.pt --tag v1

docker run -p 8000:8000 model-jacket:v1
curl -X POST http://localhost:8000/predict -H "Content-Type: application/json" -d '{"input_data": [[1,2,3]]}'

Features

• 🔌 Framework‑agnostic (Torch & ONNX out of the box)
• 🐳 One‑command containerization via jacket build
• ⚡ Low‑latency FastAPI server with async I/O
• 📈 Built‑in Prometheus metrics and /healthz endpoint
• ☸️ Production-ready Helm chart with HPA
• 🛡️ CI/CD GitHub Action for automatic build & deploy

Feel free to star ⭐, fork 🔱, and contribute! :rocket: