CUDA Engineer for AI Model Training

Apply deep GPU programming and performance optimization skills to analyze, profile, and tune GPU kernels that accelerate training and evaluation workflows for next-generation AI systems. This remote contractor role contributes technical improvements and documented findings to a customer project run in partnership with a leading AI lab. No prior experience in AI is required.

• Analyze, profile, and optimize GPU kernels using CUDA and appropriate profiling tools to maximize computational throughput on modern hardware.
• Identify kernel bottlenecks in collaboration with project stakeholders and propose targeted optimization strategies.
• Refactor C++ and CUDA codebases to improve maintainability, efficiency, and portability across diverse GPU architectures.
• Implement shader logic and graphics compute workflows using GLSL and WebGPU, ensuring integration with existing pipelines.
• Document findings, optimization steps, and measured performance improvements in clear, actionable reports.
• Participate in design discussions, evaluate new GPU-based approaches and performance metrics, and share relevant insights on advances in GPU programming.

• Required skills: CUDA, C++, GLSL, WebGPU.
• Proven experience in performance tuning GPU kernels and delivering measurable improvements.
• Advanced C++ development experience, especially in high-performance computing environments.
• Familiarity with GPU profiling tools, such as Nsight, Visual Profiler, or similar, for guided optimization.
• Strong analytical ability to reason about kernel performance across multiple hardware generations.
• Excellent written and verbal communication skills for clear documentation and technical reporting.
• Experience collaborating in remote, cross-disciplinary projects is a plus.

• Role type: Independent contractor.
• Location: Remote.
• Engagement: Contribute to a customer project focused on GPU kernel optimization in collaboration with a leading AI lab.

• Pay rate: $60.00 to $100.00 per hour.

• Must be able to work as an independent contractor.
• No prior AI experience required, domain and GPU engineering expertise are the primary qualifications.