Teach AI to solve physics problems the way a physicist would.
Leading AI labs need physicists to write challenging problems, check model derivations, and catch the sign errors and broken assumptions that still land on a plausible number. Remote, flexible, paid hourly.
Apply as an expertWhat you'll do
- Write original physics problems, from mechanics and optics to electromagnetism and thermodynamics, with worked solutions.
- Check model derivations step by step and flag where a sign, a boundary condition or a conservation law goes wrong.
- Compare and rank model responses to physics questions, and explain in writing which one holds up and why.
- Write grading rubrics and reasoning traces that show how a physicist checks units, limits and physical plausibility.
Who we're looking for
- A PhD, master's or bachelor's in physics, or current enrollment in a physics PhD program.
- Or a strong quantitative background in engineering or applied physics with years of hands-on problem solving.
- You can build a model of a physical situation, solve it, and check the result against units and limiting cases.
- You write clearly in English and can explain a flawed derivation in a few sentences.
- You use AI tools and can tell confident language from a correct answer. You will demonstrate this in the assessment.
- Location
- Remote
- Schedule
- Flexible hours
- Pay
- Hourly, set per project
- Type
- Independent contract
- To apply
- One form · about 20 minutes
How it works
- 01Tell us who you areName, email, LinkedIn and your background. No resume needed.
- 02Complete the assessmentAnswer questions on reasoning, AI skills and your field, plus two short written responses.
- 03We review your resultsIf you pass, we follow up by email to confirm your background and field.
- 04Get matched to projectsWhen a project fits your expertise, we invite you. You see the task and hourly rate before you say yes.
Why Navaday
- One form, about 20 minutesSubmit your details and assessment together. No resume upload is required.
- Show your skillsDemonstrate how you evaluate AI answers, identify errors and explain a better approach.
- Terms up frontReview the tasks, expected time commitment and hourly rate before accepting a project.
Questions
What does the assessment cover?
About 20 minutes: 3 quick reasoning questions, 5 on working with AI (checking sources, verifying an agent's claims, spotting instructions hidden in content) and 5 on AI in your field. Then two short written answers: why you picked one AI response over another, and a realistic question from your work that AI would get wrong.
How much does it pay?
The work is paid hourly. Rates depend on the project and the level of expertise it needs, so we do not post one number here. You always see the rate before you accept a project.
How much time do I need to commit?
You choose which projects to take on. Work is remote, and hours vary by project. Each invitation tells you what the work involves before you commit, so you can fit it around your job, research, or studies.
Who is the work for?
Leading AI labs and AI companies that build and test large language models. They need specialists to write challenging problems, check model reasoning, and flag answers a non-expert would accept. Client names are confidential.
What happens after I pass?
We email you about next steps. That usually means confirming your background and field, then onboarding for a specific project when one matches. Passing qualifies you for matching. It does not guarantee a project.
Can I apply from anywhere?
Right now we can place experts based in the United States, Canada and the United Kingdom. If you are somewhere else, apply anyway and we will email you when projects open in your region.
What do you do with my information?
We use your details to review your application, follow up if you do not finish it, and contact you about projects that match your field. See our privacy policy for details.
More about the assessments and what happens after you pass: expert application FAQs.