CUDA Kernel Performance Engineer

Obsidian; San Francisco (CA); On-site; USD 80,000 - 120,000; Part time

Job summary

Obsidian is seeking GPU kernel optimization experts for a contract-based project with a leading AI lab. Candidates must have strong C++ skills and practical GPU programming experience.

The role involves analyzing and optimizing GPU kernels, using profiling metrics to improve performance, and documenting decisions. Ideal candidates will be fluent in C++, familiar with CUDA or related programming models, and have at least one year of relevant experience.

Qualifications

• At least 1 year of professional or graduate-level research experience working with GPUs.
• Fluent in core C++ features through C++17.
• Strong understanding of GPU profiler performance metrics.

Responsibilities

• Analyze and optimize GPU kernels for performance, efficiency, and hardware utilization.
• Use profiler metrics to guide kernel improvements.
• Document optimization decisions clearly.

Skills

C++ programming

GPU programming

Python

CUDA

Profiling tools

Job description

Obsidian is seeking GPU kernel optimization experts for a contract-based project with a leading AI lab. Candidates must have strong C++ skills and practical GPU programming experience.

The role involves analyzing and optimizing GPU kernels, using profiling metrics to improve performance, and documenting decisions. Ideal candidates will be fluent in C++, familiar with CUDA or related programming models, and have at least one year of relevant experience.