Responsibilities
• Analyze and optimize GPU kernels for performance, efficiency, and hardware utilization
• Use profiler metrics such as L2 cache hit rate, throughput, and occupancy to guide improvements
• Review GPU kernel implementations and identify performance bottlenecks
• Write, modify, and reason about code in C++17, Python, and GPU programming models
• Apply CUDA, HIP, shader programming or related kernel techniques to enhance performance
• Document optimization decisions and explain when specific metrics are or are not useful

Requirements
• Available to work at least 20 hours per week
• Fluent in core C++ features through C++17
• Working knowledge of Python and Git
• Proficiency in at least one GPU programming model such as CUDA, HIP, Slang, HLSL, or GLSL
• At least 1 year of professional or graduate-level experience working with GPUs
• Strong understanding of GPU profiler metrics and how to apply them to kernel optimization
• Ability to optimize kernels without requiring deep algorithmic context
• Experience with CUDA, HIP, inline PTX, tensor core optimization, or NSight Compute is a plus
• Experience with NVIDIA Blackwell hardware is a plus
• Prior experience with NVIDIA, AMD, or Qualcomm is a plus
• Open-source contributions related to GPU kernel optimization are a plus

How to Apply
Click "Apply" to be taken to the Mercor website. This is not a specific job posting but an application to join their talent network. Complete your profile by uploading your resume and confirming your work location. Once verified, you will be matched to opportunities as they arise. Please note that this role cannot support H1B or STEM OPT candidates. Applying through our link supports WFH Bulletin as a referral partner, but you are welcome to apply directly if you prefer.