Senior GPU Kernel Engineer — Turbocharge Inference

relace; San Francisco (CA); On-site; USD 180,000 - 300,000; Full time

Job summary

Relace is seeking a senior, low-level systems engineer to join our San Francisco team. You will write and tune the code that runs directly on GPU hardware, pushing more work per second and per dollar while shrinking models to cut costs.

You will own performance improvements end to end, with a focus on memory, concurrency, and low-level optimization. We value demonstrable speedups and rigorous measurement, and we don’t require CUDA background—just strong fundamentals and curiosity.

Qualifications

• Five or more years of low-level performance work where the speed of the system mattered.
• Strong C or C++ and real comfort at a low level: memory, concurrency, caches.
• Track record of making something significantly faster and able to explain why.
• Habit of measuring with profiling and careful testing.

Responsibilities

• Write and tune the code that runs directly on GPU hardware.
• Push more work per second and per dollar, shrinking models to cut costs.
• Identify and fix bottlenecks; measure before and after changes.
• Own end-to-end performance improvements with a bias for rigor.

Skills

Low-level perf

C/C++

Memory

Concurrency

Profiling

Performance opt

Job description

Relace is seeking a senior, low-level systems engineer to join our San Francisco team. You will write and tune the code that runs directly on GPU hardware, pushing more work per second and per dollar while shrinking models to cut costs.

You will own performance improvements end to end, with a focus on memory, concurrency, and low-level optimization. We value demonstrable speedups and rigorous measurement, and we don’t require CUDA background—just strong fundamentals and curiosity.