Senior Software Architect - Deep Learning and HPC Communications
Core
Co-design next-gen data center platforms and scalable communications software for AI and HPC workloads, focusing on high-speed interconnects and networking.
Role type
Senior Software Architect (Deep Learning and HPC Communications)
Builds
Communication libraries (NCCL, NVSHMEM, UCX) and scalable software for large GPU clusters
Domain
High Performance Computing (HPC) and Deep Learning infrastructure
Deliverable
production ML models | infrastructure
Required skills
C/C++ programming, parallel programming models (MPI, SHMEM), communication runtimes (NCCL, NVSHMEM, UCX), Linux, system architecture, network topology algorithms
Preferred skills
CUDA programming, NVIDIA GPUs, InfiniBand/RoCE/NVLink, Deep Learning Frameworks (PyTorch, TensorFlow), HPC applications
Technologies
NCCL, NVSHMEM, UCX, MPI, SHMEM, OpenSHMEM, UCC, CUDA, PyTorch, TensorFlow, InfiniBand, RoCE, NVLink, Linux
Responsibilities
Investigate opportunities to improve communication performance by identifying bottlenecks; Design and implement new communication technologies; Explore innovative solutions in HW and SW for co-design efforts; Build proofs-of-concept, conduct experiments, and perform quantitative modeling; Use simulation to explore performance of large GPU clusters
Seniority
Senior, hands-on IC
