GitHub - 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 serve. 336x faster than Python inference.
Hacker News
March 2, 2026
AI-Generated Deep Dive Summary
Timber Ollama is a groundbreaking tool designed to revolutionize classical machine learning inference by compiling models into optimized native C99 code, delivering unprecedented speed and efficiency. Unlike traditional Python-based serving, Timber enables developers to load and serve models with just two commands, offering native latency and eliminating the overhead of the Python runtime in the inference hot path. This innovation is particularly valuable for teams seeking fast, predictable, and portable model deployment—such as fraud detection, edge/IoT applications, and regulated industries like finance and healthcare.
The tool supports a wide range of popular models, including XGBoost, LightGBM, scikit-learn, CatBoost, and ONNX. By converting these models into lightweight C99 code, Timber achieves remarkable performance improvements. For instance, it operates at an astounding 336 times faster than Python-based inference for single-sample predictions, as demonstrated in benchmarks using an Apple M2 Pro chip. This level of speed makes it ideal for low-latency transaction systems, edge deployments, and scenarios where deterministic behavior is critical.
Timber’s simplicity belies its power. It eliminates the need for complex model-serving frameworks by providing a streamlined workflow: users can load pre-trained models with `timber load
Verticals
techstartups
Originally published on Hacker News on 3/2/2026