Jan Hilgard — production AI infrastructure. Core contributor to vllm-mlx (LLM inference on Apple Silicon, quantization / KV cache). Previously founded Hosting90 (exited). https://hilgard.cz · github.com/janhilgard