I currently work on the ML accelerator performance team optimizing inference workloads on various architectures including TPUs. I was heavily involved in the MLPerf Inference benchmark suite. Research interests: Compilers, Automatic parallelization, GPUs