Local LLM inference, quantization, model evaluation, and the trade-offs between quality, speed, memory, and hardware.
Independent research in AI and security.