You will turn real physical-science and scientific-computing workflows into reproducible tasks for evaluating AI agents. The tasks will test whether an AI system can reason through complex problems, use command-line tools, debug calculations, and produce reliable files and results.
- Design multi-step terminal tasks based on realistic physics, chemistry, materials science, astronomy, and computational science workflows.
- Build self-contained computational environments with fixed software dependencies and scientific tools.
- Create input datasets, molecular structures, simulation settings, experimental data, and model configurations.
- Implement expert solutions using Python, Bash, C/C++, Julia, or other scientific programming languages.
- Create automated tests for numerical accuracy, physical consistency, convergence, output structure, units, boundary conditions, and expected scientific behavior.
- Check that workflows run consistently without downloading files during execution.
- Debug dependency, precision, solver stability, performance, and file-format problems.
What it pays and takes
This is a remote contractor assignment lasting five weeks. The work requires scientific expertise and the ability to build and validate computational workflows independently.
- Pay: $300 per task that meets the project specifications.
- Time: 20 or more hours per week.
- Schedule: Includes a four-hour overlap with Pacific Time.
- Location: Open worldwide and fully remote.
- Experience: Ph.D., postdoctoral experience, or equivalent advanced technical experience in a relevant physical-science discipline.
- Technical skills: Strong programming ability in Python, C/C++, Julia, Bash, or another scientific programming language.
- Background: Experience with Linux or terminal-based environments, numerical methods, scientific modeling, simulations, or quantitative data analysis.
- Core requirement: Ability to create and validate computational scientific workflows independently.
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site. OpenTrain is free to join and helps you find AI training work that matches your skills.
About AI training work
AI training is the human work behind modern AI systems. Contributors create examples, test model behavior, and review results so AI systems can become more accurate and reliable; specialists are paid for applying their professional knowledge to these tasks.
