Lecture by George Neville-Neil (Yale) on run-time reoptimization for heterogeneous systems, covering LLVM-based compilation, transformation, and scheduling across GPUs, TPUs, NPUs, and FPGAs.