Scientific Coding Trainer, Chemistry And Python

You will create scientific programming tasks that help train and measure AI models. Chemistry and Python are especially relevant, but tasks may cover other core STEM subjects.

Each task should turn a complex scientific idea into a clear problem with connected steps, a verified solution, and tests that show whether an AI answer is correct.

• Write one main scientific problem with at least three connected sub-problems that build toward a solution.
• Implement verified reference solutions in Python.
• Create complete unit tests and cases that distinguish correct from incorrect model outputs.
• Check each task for scientific correctness, determinism, clear wording, and the ability to measure model performance.
• Improve tasks after quality feedback and maintain a high first-pass approval standard.
• Join reviews, feedback sessions, and project standups during required overlap hours.

What it pays and takes

The rate is not provided in the listing. This is a freelance contractor assignment for an eight-week contract.

The role description calls for 40 hours per week, while the structured listing records a requirement of 20 or more hours per week. Four hours of overlap with Pacific Time are required.

• Remote work available in Bangladesh, Brazil, Colombia, Egypt, Ghana, India, Pakistan, Indonesia, Kenya, Nigeria, Turkey, and Vietnam.
• English is required for the work.
• A master's degree or PhD in chemistry, mathematics, physics, biology, computer science, or a related STEM field.
• Strong Python programming and scientific computing skills.
• Ability to write rigorous problems with clear constraints and expected outputs.
• Careful judgment about scientific correctness, deterministic results, and test quality.
• Familiarity with AI data annotation, large language model evaluation, coding benchmarks, scientific writing, or research.
• Experience with NumPy, SciPy, SymPy, or similar scientific tools.
• Published research or academic STEM project experience is helpful, especially when it shows reproducible implementation and objective evaluation of competing outputs.

How it works

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

About AI training work

AI training is the human work behind systems that learn from examples, including writing and testing code, evaluating answers, and checking whether model outputs are correct. People with strong scientific and programming skills help make those examples accurate and useful.