NVIDIA
Principal Developer, AI Networking
$272K–$489K
US, CA, Santa Clara +3 · Principal · Full-time · Posted Jun 12, 2026
- Location
- US, CA, Santa Clara
US, TX, Remote
US, CO, Remote
US, WA, Remote - Experience
- 15+ years of experience with high-performance networking.
- Education
- B.Sc in Computer Science or Software Engineering or equivalent experience.
About the job
Profiles, benchmarks, and optimizes distributed LLM training and inference workloads on large-scale GPU and CPU clusters, with a focus on collective communication and networking.
What you'll do
- Characterize workloads and analyze performance bottlenecks in distributed LLM training and inference.
- Develop PyTorch trace-based profiling, analysis, and replay tools for benchmarking, debugging, and network co-design.
- Collaborate across hardware and software teams, define performance test plans, and help achieve performance targets.
What they ask for
- B.Sc in Computer Science or Software Engineering or equivalent experience.
- 15+ years of experience with high-performance networking (RDMA, MPI, NCCL, SHARP).
- Demonstrated ability in performance evaluation techniques and approaches.
- Experience with NVIDIA GPUs and CUDA; knowledge of deep learning frameworks and collective communication libraries and protocols; proficiency in Python, Bash, and C++; experience with a container-based development environment.
Nice to have
- Extensive understanding and hands-on experience with AI workloads and benchmarking for distributed LLM training.
- Comprehensive system knowledge, including Intel / AMD / ARM CPUs, NVIDIA GPUs, HCA, Memory, PCI.
Pay
272,000 USD - 431,250 USD base salary; 320,000 USD - 488,750 USD base salary
Tools
RDMA, MPI, NCCL, SHARP, CUDA, TensorFlow, PyTorch, Python
AI focus
Characterize, benchmark, profile, and optimize distributed LLM training and inference workloads; develop trace-based tools and co-design network systems for AI workloads.
Model-written summary · the original posting is authoritative