Scientific Coding Task Trainer

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. It helps contributors discover specialized projects, build a credible AI-training profile, and grow their experience in a rapidly expanding field. Creating an OpenTrain account is free.

About AI Training and Scientific Coding

AI training is the human side of building artificial intelligence. People create examples, review outputs, and design evaluations that help modern models become more accurate and dependable. In scientific coding work, expert contributors turn complex STEM reasoning into structured programming tasks that reveal whether an AI system can produce reliable solutions.

• Work directly on training and evaluating cutting-edge AI systems
• Use scientific reasoning, programming, and structured human review
• Contribute remotely with a defined schedule and project commitment

The Scientific Coding Task Trainer Role

OpenTrain is hiring a Scientific Coding Task Trainer to create high-quality scientific programming tasks for training and evaluating frontier AI models. You will translate challenging concepts into rigorous, well-posed coding problems, implement verified Python solutions, and build tests that distinguish reliable model responses from incorrect ones.

The work spans mathematics, physics, chemistry, biology, computer science, and related STEM disciplines. It combines advanced scientific reasoning, software implementation, and careful evaluation of model performance.

• Role focus: Scientific Coding Task Evaluation
• Experience level: Entry level
• Engagement: Remote contractor assignment
• Commitment: 40 hours per week for eight weeks
• Overlap requirement: Four hours with Pacific Time
• Compensation: Undisclosed

What You’ll Do

You will be responsible for developing complete, testable scientific coding tasks and improving them based on feedback and model evaluation results. Strong first-submission quality and active participation in project communication are important parts of the assignment.

• Write a main scientific problem with at least three logically connected sub-problems leading to a final solution
• Implement verified golden solutions in Python
• Provide complete unit test coverage for each solution
• Design discriminative test cases that separate correct and incorrect outputs
• Validate task well-posedness, scientific correctness, determinism, and test-case quality against defined rubrics
• Refine tasks in response to quality feedback and language-model evaluation results
• Participate in review discussions, feedback sessions, and project standups during overlap hours

Required Qualifications

A master’s degree or PhD in mathematics, physics, chemistry, biology, computer science, or a related STEM discipline is required. You should also bring strong Python programming and scientific-computing ability, along with the judgment needed to assess scientific correctness, determinism, and evaluation quality.

• Master’s or PhD in a relevant STEM discipline
• Strong Python programming and scientific-computing skills
• Experience implementing unit tests
• Ability to formulate rigorous, constrained scientific problems with clear expected outputs
• Ability to judge whether test cases distinguish correct from incorrect model responses
• Careful evaluation of scientific correctness and deterministic task behavior

Helpful Background

The following experience is useful but is presented as helpful background rather than a required qualification. It can support your ability to design realistic tasks and evaluate AI-generated code.

• Prior AI data annotation or scientific-writing experience
• Familiarity with LLM evaluation frameworks or coding benchmarks
• Experience with NumPy, SciPy, SymPy, or comparable scientific tools
• Published research or academic project experience in a STEM field

Location and Schedule

This remote contractor assignment is available to candidates located in Bangladesh, Brazil, Colombia, Egypt, Ghana, India, Pakistan, Indonesia, Kenya, Nigeria, Turkey, or Vietnam. The project requires 40 hours per week, including four hours of overlap with Pacific Time, for an eight-week contract.

• Eligible countries: Bangladesh, Brazil, Colombia, Egypt, Ghana, India, Pakistan, Indonesia, Kenya, Nigeria, Turkey, and Vietnam
• Language: English
• Work arrangement: Remote
• Contract length: Eight weeks
• Weekly commitment: 40 hours
• Pacific Time overlap: Four hours

Build Your AI Training Career with OpenTrain

AI training and data-labeling work can help specialists apply their expertise to the systems shaping modern technology. Through OpenTrain, you can build a lasting portfolio of relevant work, show credible experience, and find projects aligned with your scientific and programming skills.

• Create an OpenTrain account for free
• Build a profile that reflects your AI training experience
• Apply to specialized projects in minutes
• Develop a portfolio across the growing AI-training industry