Python Domain Expert CUDA C++ AI Code Evaluation

You will help improve large language models through coding prompt development, training-data creation, and technical evaluation. Your main focus will be translating CUDA and C++ implementations into equivalent Python code with PyTorch and NumPy while preserving their logic, numerical behavior, readability, and performance.

• Translate CUDA and C++ implementations into Python with PyTorch and NumPy.
• Analyze CUDA kernels, GPU-accelerated algorithms, code structure, efficiency, and function.
• Review generated code translations, debug defects, and provide precise corrective feedback.
• Create prompts based on real-world CUDA and PyTorch programming tasks.
• Rank alternative model responses and explain your technical decisions.
• Document coding decisions and suggest improvements based on recurring model failures.

This is a fully remote, three-month contractor assignment. Pay details are not provided.

• Work commitment: At least 20 hours per week and at least 4 hours per day.
• Schedule options: 20, 30, or 40 hours per week, including 4 hours of overlap with Pacific Standard Time.
• Location: Open to candidates in India, Pakistan, Nigeria, Kenya, Egypt, Ghana, Bangladesh, Turkey, Brazil, and Mexico.
• Language: Professional fluency in written and spoken English.
• Experience: At least five years of overall professional experience, including three or more years with Python and two or more years with CUDA and C++.
• Technical skills: Strong Python skills for scientific computing with PyTorch and NumPy.
• Technical knowledge: CUDA programming, C++ fundamentals, GPU computation, parallelism, numerical consistency, and debugging.
• Communication: Ability to explain technical feedback clearly in writing and speech.
• Helpful background: Experience evaluating AI-generated code or contributing to large language model tuning workflows.

Apply on OpenTrain with your resume, then complete the application on the hiring site.

AI training is the human work behind artificial intelligence, including writing prompts, reviewing model responses, and checking whether generated code works correctly. People with strong technical experience are needed to spot subtle errors and give clear feedback that helps models produce more reliable results.