Generating GPU kernels automatically to speed up ML models. Ex-Intel, worked on CPU microcode and ML accelerators.
Luminal compiles AI models for optimized GPU and ASIC inference.