Senior Systems Performance Engineer

Description

The company has been building graphics, PC gaming and accelerated computing technology for over 25 years, and it is now focused on AI and the next generation of computing, where GPUs run computers, robots and self-driving cars.

The team is a group of hardware and software engineers in the DGX Server Product Engineering group, building and running complex automated test plans for its GPU accelerated computing servers.

This is a senior validation role for someone who wants to work on hard problems alongside strong people.

Responsibilities

• Write and tune synthetic GPU workloads in CUDA or GPU programming DSLs, and build application workload recipes on AI training and inference frameworks, to stress specific hardware subsystems and widen test coverage.
• Bring current AI training and inference workloads up on new GPU platforms.
• Use architecture analysis and system-level performance models to describe how workloads behave, locate bottlenecks, and decide which workloads to run and what they should stress.
• Run controlled experiments and compare model predictions with workload profiles, system telemetry and measured results, then refine workloads, models and test strategies where they disagree.
• Investigate performance anomalies, workload correctness problems, and hardware and software failures, build reproducible cases, and work with architecture, design and software teams to get fixes made and confirm them.
• Build automation covering workload deployment, experiment runs, telemetry collection and analysis across systems and configurations.

Requirements

• A BS in Computer Engineering, Electrical Engineering, Computer Science or a related field, or equivalent experience.
• 5+ years of experience in systems performance engineering, performance modeling, accelerated computing, or complex system development and debugging.
• A deep understanding of computer architecture, including GPU and CPU execution, memory hierarchies and interconnects, and how each affects application performance.
• Experience building analytical or simulation-based performance models and checking their predictions against measurements on running systems.
• Hands-on experience deploying, profiling and tuning AI training or inference workloads with frameworks such as PyTorch, TensorRT-LLM, vLLM or SGLang.
• Strong Python and data analysis skills, including designing controlled experiments, reading telemetry and explaining findings clearly.
• Working knowledge of CUDA and GPU profiling tools, and the ability to link application behavior to hardware utilization and bottlenecks.
• Experience debugging workloads and system behavior on Linux, and working across hardware and software teams.
• Nice to have: experience with the company Dynamo, TensorRT-LLM, vLLM, SGLang or NCCL.
• Nice to have: CUDA kernel development or optimization using cuBLAS, CUTLASS or similar libraries.
• Nice to have: familiarity with x86 or Arm servers, NVLink, PCIe, InfiniBand or Ethernet, and distributed GPU systems.
• Nice to have: experience using hardware error, power, thermal and reliability telemetry to build targeted stress coverage and isolate failures.
• Nice to have: a record of building automation and turning experimental results into measurable gains in performance, coverage or product quality.

Conditions

Base salary depends on location, experience and the pay of people in similar roles. The base range is 136,000 USD to 212,750 USD for Level 3, and 168,000 USD to 258,750 USD for Level 4.

Applications are accepted at least until November 1, 2026.

This posting is for an existing vacancy.

The company uses AI tools in parts of its recruiting process.

The company is an equal opportunity employer and does not discriminate on the basis of any legally protected characteristic.

Benefits

• Highly competitive salaries and a comprehensive benefits package, with details available on the company's benefits website.