Manager, Software Engineering - NCCL
Core
Lead the China NCCL team to develop high-performance communication libraries (NCCL, NVSHMEM) for Deep Learning and HPC applications running on tens of thousands of GPUs.
Role type
Manager, Software Engineering (GPU Communications Libraries)
Builds
NCCL and NVSHMEM communication runtimes for DL and HPC
Domain
High Performance Computing (HPC) and Deep Learning (DL)
Deliverable
production ML models | infrastructure
Required skills
Systems software, communication runtimes, high performance networking, computer systems architecture, networking technologies (RDMA, RoCE, Ethernet, InfiniBand), operating systems principles, HW-SW interactions, performance analysis/optimizations, C/C++ programming, Linux debugging
Preferred skills
Parallel programming models (MPI, SHMEM), communication runtimes (NCCL, NVSHMEM, UCX), CUDA, MPI, OpenMP, OpenACC, pthreads, Deep Learning Frameworks (PyTorch, TensorFlow, vLLM, SGLang, TRT-LLM)
Responsibilities
Lead, mentor, and grow the China engineering team; Own end-to-end execution spanning planning, prioritization, quality control, and performance; Interact with customers and researchers to define product roadmap; Contribute to feature design and implementation; Review and improve processes, infrastructure, and practices
Seniority
Manager, 4+ years management experience
