kossisoroyce/timber: Ollama for classical ML models. AOT compiler that turns XGBoost, LightGBM, scikit-learn, CatBoost & ONNX models into native C99 inference code. One command to load, one command to

Timber is an AOT compiler that converts trained tree-based ML models (XGBoost, LightGBM, scikit-learn, CatBoost, ONNX) into optimized native C99 code and serves them via a local HTTP API. It eliminates the Python runtime from the inference path, achieving microsecond latency (~2 µs native call vs. hundreds of µs with Python). The workflow mirrors Ollama: one command to load a model, one to serve it. Target use cases include fraud detection, edge/IoT deployments, regulated industries needing deterministic artifacts, and platform teams replacing Python serving overhead. Benchmarks show up to 336× speedup over Python XGBoost single-sample inference on Apple M2 Pro hardware.