The work
You will create and solve difficult physics problems used to evaluate and improve large language models. Your work will help test model reasoning across advanced physical sciences, from early undergraduate topics through graduate and PhD-level material.
Because the partner title identifies this as a Mathematics Expert role while the duties focus on physics-based evaluation, review the listed duties and qualifications carefully before applying.
- Design original physics problems that test reasoning and reveal model limitations.
- Solve problems across undergraduate, graduate, and PhD-level curricula.
- Write clear, detailed, step-by-step solutions that can serve as strong reference examples.
- Help define evaluation benchmarks for symbolic manipulation, abstraction, and multi-step reasoning.
- Provide annotations and feedback on complex reasoning paths.
- Work with language model researchers to align tasks with evaluation goals.
What it takes
This is a remote, part-time contract role open worldwide. The work requires strong mathematical knowledge, clear written communication, and the ability to assess complex reasoning independently.
- Commitment of 20 or more hours per week.
- Fluency in English comprehension and written communication.
- Strong mathematics foundation from advanced entrance-level concepts through graduate or PhD-level material.
- Strong analytical, research, and problem-solving skills.
- Ability to develop creative scenarios and explain complex ideas in a structured way.
- Comfort working independently in a remote setting.
- A personal desktop or laptop with stable, high-speed internet.
- A degree or doctorate in mathematics, or an equivalent technical field, is expected.
- Previous experience in AI evaluation, data annotation, content review, quality assurance, or a related analytical role is helpful but not required.
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 modern AI systems, including writing examples, rating model responses, and evaluating reasoning. People with strong subject knowledge help create accurate examples and judge difficult outputs, which is why specialized expertise matters in this field.