Yu Shao#
Yu Shao works on AMD’s AI Models team as technical leader, focused on model enablement and benchmarking automation for AI training and inference workloads on AMD Instinct GPUs single node across multi node distributed setups of SLURM and Kubernetes. That work centers on MAD and madengine, the declarative model registry and execution engine that turn LLMs enablement into reproducible, comparable performance measurements across inference serving stacks such as vLLM, SGLang, ATOM, and P/D disaggregated frameworks, and training frameworks such as AMD’s Primus, TorchTitan, Megatron-LM, JAX MaxText, and so on.
Posts by Yu Shao
