AI & ML interests

Building production-quality ML inference in native code — ONNX graph construction, hand-rolled serialization, and model export tooling. Focused on making voice AI (RVC, HuBERT, VITS, vocal separation, forced alignment) run fast without Python or libtorch.