I'm a core OpenXLA contributor at Google. Leading a small but mighty team that makes sure that XLA works great on modern CPUs (x86 and ARM) and serves millions of QPS at Google. Also work on XLA:GPU to make it fastest ML compiler on latest NVIDIA GPUs and make it utilize all the latest and greatest features CUDA has to offer. Building heterogeneous TPU/GPU + CPU compiler for mixing device-acceleterated and host computations, all from a single JAX program!






