Ryan Shamim

Efficient LLM inference, GPU acceleration, structured reuse, advanced attention kernels, symbolic compute, and hybrid neuro-symbolic inference.