Show HN: Furnace – Rust and Burn inference server, zero Python, single binary

4 points

a year ago

*Furnace* is a high‑performance ML inference server written in pure Rust, powered by the Burn framework.

* Key Features* - Zero Python runtime — everything runs as a single Rust binary (~2.3 MB) - Sub‑millisecond inference time (~0.5 ms on MNIST‑like models) - REST HTTP API: `/predict`, `/healthz`, `/model/info` - Production‑ready: graceful shutdown, error handling, CORS support

* Quick Start* ```bash git clone https://github.com/Gilfeather/furnace.git cd furnace cargo build --release ./target/release/furnace --model-path ./sample_model --port 3000

# Health check curl http://localhost:3000/healthz

# Model info curl http://localhost:3000/model/info

# Inference curl -X POST http://localhost:3000/predict \ -H "Content-Type: application/json" \ -d '{"input":[0.1,...,0.1]}'