Computational Physics Problem Designer (Python)

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. We connect experts to project-based work that helps shape state-of-the-art AI systems.

This role is part of OpenTrain’s mission to grow the human side of AI: contributors prepare high-quality examples and solutions that models learn from, while building a flexible, remote career.

Why AI training work matters

AI models learn from examples and human feedback. Designing verified, research-style computational physics problems directly improves model performance for scientific and educational tasks.

These projects are well-suited to people who want cutting-edge, flexible remote work that leverages domain expertise and programming skills to influence how models reason about physics.

• 100% remote, flexible hours — fit work around studies, a job, or family.
• Contribute to the real-world datasets and evaluations used to train and fine-tune AI.

The role

We are seeking an experienced computational physicist to design and fully verify research-style problems and reproducible solutions in Python. Projects are part-time, contract-based, and focus on problems representative of real research workflows.

Work includes authoring clear problem statements, producing code and numerical tests using scientific Python libraries, and responding to QA feedback to meet annotation guidelines.

• Employment type: Contractor, Part-time.
• Time requirement: 20+ hours/week.
• Pay: USD $15–$60 per hour (listed hourly rate: $60/hr; final rate depends on scope and complexity).

What you'll do

Create original computational physics problems that reflect research workflows and target areas such as mechanics, electromagnetism, thermodynamics, and quantum mechanics.

Provide fully verified, reproducible solutions and well-documented code using scientific Python libraries so outputs can be used for model training, evaluation, or fine-tuning.

• Write clear, self-contained problem statements and expected outputs.
• Implement and test solutions with NumPy, SciPy, SymPy, and other scientific Python tools.
• Use numerical simulation techniques (e.g., numerical integration, Monte Carlo) where appropriate.
• Follow detailed annotation guidelines and iterate on QA feedback from reviewers.

Requirements

All substantive requirements below are mandatory unless noted otherwise. You must be able to produce reproducible Python code and clear written solutions in English.

Submit your CV in English and indicate your level of English proficiency; include an email address and phone number on your CV.

• Bachelor’s degree or higher in Physics or a closely related field.
• 2+ years of applied, research, or teaching experience in physics.
• Advanced Python skills with hands-on experience using NumPy, SciPy, SymPy, etc.
• Experience with numerical simulation methods (integration, Monte Carlo, ODE solvers, etc.).
• Hands-on text annotation or review experience and familiarity with following annotation guidelines.
• Strong written and spoken English; professional written communication skills.
• Ability to accept and act on QA feedback and produce reproducible solutions.

Who should apply

You are a computational physicist, researcher, or instructor who enjoys turning research problems into clear, testable exercises with production-quality Python code.

This role suits people who want steady part-time, remote contract work and who can commit to 20+ hours per week while delivering reproducible solutions that meet strict QA.

Locations, how it works, and application

This position is open globally but excludes certain locations. Applicants must confirm they are not based in any restricted locations listed below.

To apply, send your CV (in English) with level of English proficiency, email, and phone number. Shortlisted candidates will be asked to complete a short paid sample task to demonstrate problem-design and reproducibility skills.

• Restricted locations (applicants cannot be based in): Iran, Cuba, North Korea, Syria, Sudan, Venezuela, Myanmar, Russia, Belarus, Palestine, Switzerland, China, Taiwan, Kenya; U.S. states: Alaska, Arkansas, California, Connecticut, Delaware, Georgia, Hawaii, Illinois, Indiana, Kansas, Louisiana, Mai
• Labeling task types include text generation, evaluation/rating, and fine-tuning preparation; labeling software is 'OTHER' and the role requires producing code and annotated text outputs.