Senior GPU Kernel Performance Engineer - Hybrid, Bellevue

Designworks Talent; Bellevue (WA); Hybrid; USD 180,000 - 240,000; Full time

Benefits offered by this job

Health insurance

401(k) plan

Annual bonus

Long term incentives

Job summary

Designworks Talent is seeking a GPU Performance / Kernel Engineer in a hybrid Bellevue, WA setting. The role focuses on optimizing GPU kernels, reducing latency, and maximizing throughput for AI workloads.

Ideal candidates will have deep GPU architecture knowledge, experience with CUDA/ROCm, and a track record of performance engineering in large-scale systems. Join a lean, collaborative team building high-performance AI infrastructure.

Qualifications

• Experience with GPU kernel development and performance optimization (CUDA/ROCm).
• Proven ability to improve GPU utilization and reduce latency for production AI workloads.
• Deep understanding of GPU architecture, memory hierarchy, parallel computing.
• Experience profiling and debugging performance issues in AI/distributed environments.
• Ability to own technically complex problems in a fast-moving team.
• Strong systems programming and performance engineering mindset.

Responsibilities

• Profile, analyze, and optimize GPU kernels to improve latency and throughput.
• Identify and eliminate data-plane bottlenecks across large-scale AI workloads.
• Tune performance-critical workloads for training and inference.
• Collaborate with AI infrastructure, ML, and platform engineering teams.
• Develop benchmarking methodologies and performance measurement practices.
• Evaluate emerging GPU technologies and optimization techniques.
• Contribute to practices that improve GPU efficiency and reliability.

Skills

GPU kernel development

CUDA

ROCm

Performance optimization

Profiling

Systems programming

Tools

Nsight Systems

Nsight Compute

ROCm profiling tools

Job description

Designworks Talent is seeking a GPU Performance / Kernel Engineer in a hybrid Bellevue, WA setting. The role focuses on optimizing GPU kernels, reducing latency, and maximizing throughput for AI workloads.

Ideal candidates will have deep GPU architecture knowledge, experience with CUDA/ROCm, and a track record of performance engineering in large-scale systems. Join a lean, collaborative team building high-performance AI infrastructure.