GPU Programming Software Engineer [Remote]

Role Overview

Apply GPU programming expertise to create tasks that help train large language models, with an emphasis on performance and architecture.

Key Responsibilities

• Design GPU-related training tasks for large language models.
• Design, implement, and optimize GPU-based software using CUDA, WebGPU, or GLSL.
• Profile and tune GPU kernels and shaders for performance and efficiency.
• Develop host-side logic and integrations in C++.

Qualifications

• Advanced experience with a GPU programming language that runs on NVIDIA chips.
• Proficiency in C++.
• Experience with CUDA, WebGPU, or GLSL.
• Background in graphics programming, ML acceleration, scientific computing, high-performance computing, or another GPU-focused domain.

Work Terms

• Remote, independent contractor engagement.
• Compensation is output-based and paid for each task that meets project specifications.
• Completion time may vary based on experience and workflow.
• Minimum submission requirements apply, including a required minimum number of task submissions each week.

Compensation

$60 to $95 per hour.

Availability and Application Process

Roles are typically filled within 48 hours, and candidates should be ready to begin promptly. If selected, you will be expected to begin your first tasks within 24 to 48 hours after completing onboarding.