Physics Problem Evaluator

Claru; United States; On-site; USD 76,000 - 138,000; Full time

Job summary

AI Trainer Jobs in the United States seeks a graduate-level physicist to review AI-generated solutions to university- and graduate-level physics problems across mechanics, electromagnetism, thermodynamics, and beyond. You will read a model's step-by-step reasoning, identify errors, and rate the output on correctness, coherence, and depth of explanation.

Your work includes articulating precisely where the reasoning broke down and, when asked, providing a corrected reference solution.

Qualifications

• Graduate-level background in physics or closely related field.
• Professor- or PI-level experience strongly preferred.

Responsibilities

• Review AI-generated solutions to physics problems and identify errors.
• Articulate precisely where reasoning broke down and provide corrected solutions upon request.
• Log in, claim a batch, complete it, and submit via the web interface.

Skills

Physics problem solving

Graduate-level mechanics and electroma

Technical written communication

Critical evaluation of derivations

Meta-level reasoning

Education

Graduate-level physics degree

Job description

Pay: $55–100/hour

You'll review AI-generated solutions to university-level and graduate physics problems — mechanics, electromagnetism, thermodynamics, quantum mechanics, and beyond. Each task involves reading a model's step-by-step response, identifying errors in reasoning or calculation, and rating the output on correctness, logical coherence, and depth of explanation. You won't just flag wrong answers; you'll articulate precisely where the model's reasoning broke down and, where asked, provide a corrected reference solution.

Your work spans three main workflows: ambiguous-case judgment (deciding when a non-standard approach is still valid), written technical feedback (annotating specific failure modes in a model's derivation), and problem-solving depth review (assessing whether a solution demonstrates genuine physical intuition or is just symbol-shuffling). Tasks arrive in batches via a structured web interface; you log in, claim a batch, complete it, and submit — no scheduling calls, no meetings.

The role requires at minimum a graduate-level background in physics or a closely related field. Professor- or PI-level experience is strongly preferred because many tasks involve research-adjacent material: statistical mechanics, condensed matter, or computational physics at the level of a first-year PhD course. You'll be compensated at the Expert tier and are expected to maintain accuracy above the project's threshold, which will be shared during onboarding.

Skills & requirements

• Physics problem solving
• Graduate-level mechanics and electromagnetism
• Technical written communication
• Critical evaluation of mathematical derivations
• Meta-level scientific reasoning

Category: RLHF