GPU / AI Infrastructure Engineer @ Quant Trading Firm - Hong Kong / Shanghai - 3+ years exp

About the role

Build and operate large-scale GPU clusters, including scheduling, networking, storage, and reliability, and optimize performance across GPUs, CPUs, memory, and interconnects. Develop AI training and inference platforms, distributed training systems, model-serving tools, and high-throughput HPC infrastructure for research teams. Candidates should have strong C++ and/or Python skills and hands-on experience in GPU computing, HPC, or low-level systems programming, along with solid Linux, operating systems, and computer architecture fundamentals. The posting seeks approximately 3–6 years of experience, with exceptional PhD candidates with 2–3 years also considered; relevant technologies include CUDA, Triton, NCCL, RDMA/InfiniBand, Kubernetes or Slurm, distributed training, and inference engines.