I'm a Rust engineer and maintainer of Burn, an open-source deep learning framework written in Rust.

I work on speech recognition and ML inference, building fast, portable systems that run everywhere from servers to embedded devices.

• Burn: deep learning framework in Rust. I contribute across the stack: ops, backends, ONNX import, WASM, no_std, and performance.
• burn-onnx: ONNX model importer for Burn with broad model coverage, codegen, and GPU/CPU correctness.
• burn-store: model weights import/export in Burn with zero-copy loading, quantized tensor storage, and mixed-precision support.
• burn-flex: fast, memory-efficient CPU backend for Burn. Multi-threading, SIMD, and optimized matmul. Runs on std, no_std, and WASM with f16/bf16 and zero-copy loading.
• wordchipper: HPC BPE tokenizer in Rust, up to 9x faster than tiktoken, with no_std, WASM, and Python bindings.
• Speech recognition: applying Burn to real-world audio and ASR pipelines.
• hstats: lightweight online histograms and stats for data streams, no_std and parallel-friendly.

• Systems programming in Rust (no_std, SIMD, WASM)
• ML inference and model portability
• Deep learning framework internals
• Speech and audio processing