Myles Doyle

Making models train quickly and execute inference as fast as possible on different hardware architectures (specifically NPUs and GPUs).