Staff GPU Performance Engineer – Production Inference

Speedrun Talent Network; San Francisco (CA); On-site; USD 210,000 - 360,000; Full time

Job summary

Gimlet seeks a Member of Technical Staff to drive performance for production AI inference across a heterogeneous accelerator fleet. You will analyze kernel execution, memory behavior, and scheduling, and connect hardware signals to end-to-end latency and throughput.

You will build measurement systems, run controlled experiments, and push improvements across kernels, compilers, runtimes, and distributed execution. The role partners with multiple engineering teams to scale performance.

Qualifications

• Experience profiling GPU/accelerator workloads and turning measurements into production improvements.
• Strong fundamentals in software engineering for performance-critical systems.
• Bachelor's degree in a relevant field or equivalent experience.

Responsibilities

• Characterize and improve performance of production AI inference across heterogeneous accelerators.
• Design controlled experiments to identify bottlenecks in compute, memory, and scheduling.
• Build measurement systems and methodologies for end-to-end latency, throughput, and cost.
• Translate findings into kernel, compiler, and runtime optimizations.
• Develop repeatable processes for evaluating new accelerators within Gimlet's platform.

Skills

Performance profiling

GPU execution understanding

Controlled experiments design

Software engineering fundamentals

Production systems experience

Education

Bachelor's degree

Tools

Nsight Systems

Nsight Compute

Perfetto

Job description

Gimlet seeks a Member of Technical Staff to drive performance for production AI inference across a heterogeneous accelerator fleet. You will analyze kernel execution, memory behavior, and scheduling, and connect hardware signals to end-to-end latency and throughput.

You will build measurement systems, run controlled experiments, and push improvements across kernels, compilers, runtimes, and distributed execution. The role partners with multiple engineering teams to scale performance.