Senior GPU/CPU Profiling Architect

NVIDIA Corporation; Durham (NC); Hybrid; USD 184,000 - 357,000; Full time

Job summary

NVIDIA Corporation seeks a Senior Systems Performance Engineer to design and optimize profiling infrastructure across CPU and GPU systems. You will measure, analyze, and optimize interactions in multi-core/sockets and NVLink environments, focusing on high-throughput GPU workloads and EDA applications.

The role requires deep knowledge of CUDA and profiling tools (Nsight, perf, VTune, eBPF, Valgrind) and experience with distributed schedulers (Slurm, LSF, Kubernetes).

Qualifications

• CUDA programming model is essential and experience with GPU profiling tools to address bottlenecks.
• Extensive knowledge of profiling tools such as perf, eBPF, VTune, or Valgrind.
• Ability to distill complex performance data into actionable roadmaps.
• Experience with distributed compute environments (Slurm, LSF, or Kubernetes).
• BS/MS/PhD in CS/EE or related field with 8+ years in systems-level performance analysis.

Responsibilities

• Architecting and maintaining custom profiling frameworks across CPU and GPU environments.
• Benchmarking EDA apps to characterize memory access, cache hit rates, and ILP.
• Using GPU profilers to detect warp divergence, occupancy, and PCIe/NVLink bottlenecks.
• Developing tools to monitor high-watermark memory usage in large EDA builds.
• Developing predictive models for hardware procurement and cloud instance selection.

Skills

CUDA profiling

Nsight Tools

perf

eBPF

VTune

Valgrind

Kubernetes

Slurm

GPU/CPU profiling

Education

BS/MS/PhD in Computer Science or Electrical Engineering

Tools

Nsight Systems/Compute

perf

eBPF

VTune

Valgrind

Kubernetes

Slurm

LSF

Job description

NVIDIA Corporation seeks a Senior Systems Performance Engineer to design and optimize profiling infrastructure across CPU and GPU systems. You will measure, analyze, and optimize interactions in multi-core/sockets and NVLink environments, focusing on high-throughput GPU workloads and EDA applications.

The role requires deep knowledge of CUDA and profiling tools (Nsight, perf, VTune, eBPF, Valgrind) and experience with distributed schedulers (Slurm, LSF, Kubernetes).