Gimlet Labs
Member of Technical Staff - GPU Performance
$150K–$390K
San Francisco, CA · On-site · Staff · Full-time · Posted Oct 26, 2026
- Education
- Bachelor’s degree in a relevant field, or an equivalent combination of education, training, and professional experience.
About the job
Characterize and improve production AI inference performance across Gimlet’s heterogeneous accelerator fleet, translating hardware-level analysis into end-to-end production optimizations.
What you'll do
- Establish performance baselines, diagnose workload bottlenecks, and evaluate new accelerators.
- Optimize execution across kernels, runtimes, memory, synchronization, scheduling, and multi-device workloads.
- Build automated benchmarks and regression detection, and partner with engineering teams to deliver production improvements.
What they ask for
- Strong software engineering fundamentals and experience with performance-critical systems.
- Experience profiling GPU or accelerator workloads and applying findings to production improvements.
- Strong understanding of GPU execution, memory hierarchies, asynchronous execution, synchronization, and performance tradeoffs.
- Bachelor’s degree in a relevant field, or an equivalent combination of education, training, and professional experience.
Nice to have
- Experience with GPU profiling tools, hardware counters, or accelerator programming models.
- Familiarity with inference optimization, performance analysis, distributed execution, or benchmark infrastructure.
Pay
$150K – $390K USD base salary
Tools
Nsight Systems, Nsight Compute, rocprof, Omniperf, VTune, Perfetto, CUDA, HIP
AI focus
Improves performance of production AI inference across heterogeneous accelerators.
Model-written summary · the original posting is authoritative