Senior Deep Learning Framework Communications Engineer
Core
Integrate communication libraries and optimize multi-GPU communication for AI frameworks like PyTorch, TRT-LLM, and vLLM to support scaling from microsecond latency inference to 100K GPU training clusters.
Role type
Senior Deep Learning Framework Communications Engineer
Builds
AI toolkits, communication libraries (NCCL, NVSHMEM), and high-performance AI stacks
Domain
Artificial Intelligence, High Performance Computing, Distributed Systems
Deliverable
production ML models | infrastructure
Required skills
Deep Learning Framework integration, C++/CUDA development, AI workload analysis, AI compiler optimization, multi-GPU performance benchmarking, HPC communication concepts, fault-tolerant system design, kernel authoring
Preferred skills
Parallel programming on NCCL/NVSHMEM/MPI, HW-SW interaction knowledge, AI compiler pattern matching, memory hierarchy understanding, compute-communication overlap programming
Technologies
PyTorch, JAX, TRT-LLM, vLLM, SGLang, NCCL, NVSHMEM, CUDA, Triton, cuTe, Python, Nsight Systems, torch.compile
Responsibilities
Integrate new communication library features from PoC to production, analyze AI workloads for multi-GPU communication requirements, improve AI compilers for communication hiding/fusion, characterize performance on multi-GPU clusters, design fault-tolerant elastic solutions, author custom communication kernels, influence communication library roadmaps
Seniority
Senior, hands-on IC