Haidong Chen — GPU AI Factory Principal Engineer

I design, commission, and tune large-scale NVIDIA GPU AI factories: rack-scale architecture, InfiniBand and Spectrum-X fabrics, direct liquid cooling, schedulers, and training recipes for Blackwell and Hopper systems.

AI factory engineering expertise

• GB200 and GB300 NVL72 rack-scale architecture, cluster bring-up, and acceptance testing
• InfiniBand NDR/XDR, Spectrum-X, NCCL benchmarking, and rail-optimised network fabrics
• Direct liquid cooling, CDU commissioning, thermal telemetry, and facility integration
• FP8 and NVFP4 model tuning with NeMo, Megatron-Core, TensorRT-LLM, vLLM, and SGLang

Explore

Read my professional profile, publications, technical talks and articles, and robotics simulations.