AI网络通信库研发工程师（J97232）
Core
Develop and optimize AI communication libraries to resolve network throughput and latency issues for customer model training and inference on public clouds.
Role type
Senior IC AI infrastructure engineer (communication libraries)
Builds
High-performance AI communication and collective communication libraries
Domain
Cloud computing + AI infrastructure
Deliverable
production ML models
Required skills
C++, Python, CUDA, MPI, OpenMP, performance analysis tools (NVIDIA Nsight Systems), RDMA, GPUDirect RDMA, NVLink, NCCL, UCX, NVSHMEM
Preferred skills
Experience with large model training and inference scenarios, deep understanding of communication protocols and data processing bottlenecks
Responsibilities
Debug and optimize network performance for model training and inference; Analyze and fix bottlenecks in communication protocols and data processing; Evaluate feasibility of latest large model technologies for public cloud applications
