Senior AI Networking Architect for High-Perf GPU Clusters

NVIDIA Corporation; United States; Sur place; USD 272 000 - 489 000; Plein temps

Avantages offerts par ce poste

Equity

Benefits

Résumé du poste

NVIDIA is seeking a senior software engineer for AI Networking Codesign and Benchmarking R&D. You will profile, optimize AI workloads on large-scale GPU/CPU clusters, focusing on distributed systems and high-performance networking, using NCCL and CUDA.

Responsibilities include benchmarking, profiling, and developing trace-based tools, collaborating across teams, and defining performance targets with equity and benefits offered.

Qualifications

• B.Sc in Computer Science or Software Engineering or equivalent experience.
• 15+ years of experience with high-performance networking (RDMA, MPI, NCCL, SHARP).
• Experience with NVIDIA GPUs and CUDA; PyTorch or TensorFlow knowledge.
• Strong analytical and problem-solving abilities.

Responsabilités

• Characterize AI workloads and distributed LLM training/inference on NVIDIA systems.
• Benchmark, profile, and analyze performance focusing on networking.
• Develop PyTorch trace-based profiling and replay tools for benchmarking.
• Collaborate across hardware to software teams for insights.
• Define performance test plans and target achievement.

Connaissances

High-performance networking

Python

C++

Bash

NCCL

RDMA

CUDA

Formation

Bachelor's degree in CS or SWE or equivalent

Outils

CUDA

NCCL

Description du poste

NVIDIA is seeking a senior software engineer for AI Networking Codesign and Benchmarking R&D. You will profile, optimize AI workloads on large-scale GPU/CPU clusters, focusing on distributed systems and high-performance networking, using NCCL and CUDA.

Responsibilities include benchmarking, profiling, and developing trace-based tools, collaborating across teams, and defining performance targets with equity and benefits offered.