Principal Software Engineer, E2E Performance and Goodput — CSP Engagements

Job Description

Lead end-to-end performance characterization and validation with CSP and hyperscale customers, ensuring workloads meet performance targets on NVIDIA platforms. Analyze cross-customer performance patterns, incorporate customer feedback into optimization priorities, and coordinate improvements to tools, configurations, and NVIDIA software and hardware systems.

Requirements: Requires 15 or more years of systems performance engineering experience, ideally in GPU, HPC, or ML infrastructure, and a bachelor's or master's degree in a related field or equivalent experience. Candidates should have GPU profiling and distributed training expertise, strong statistical and data analysis skills, knowledge of the software stack's performance effects, and the ability to influence engineering teams and communicate with technical and executive audiences.

Key Skills: Systems Performance Engineering, GPU Workload Profiling, Distributed Training Performance, Performance Characterization, Performance Analysis, Statistical Analysis, Python, Pandas, Performance Regression Detection, CUDA, NCCL, GPU Architecture, Memory Bandwidth Optimization, Customer Engagement, Cross-Functional Leadership, Benchmarking

Benefits: Equity, Benefits