Physics Specialist (LLM Evaluation)

As a Physics Specialist, you will help improve large language models in specialized physics topics. You will write prompts, review model answers, and use your physics knowledge to identify accurate explanations and reasoning errors.

The work is remote and primarily asynchronous. Assignments become available throughout a year-round program, but project availability varies by subject area.

• Create prompts that explore physics concepts, topics, and subdomains.
• Review large language model responses for conceptual accuracy, relevance, quality, and reasoning.
• Research physics topics with assistance from AI.
• Explain and assess physics reasoning to distinguish expert understanding from model-generated errors.
• Work independently on AI research assignments as they become available.

This is flexible, part-time contractor work for physics specialists.

• Pay: up to $50 per hour.
• Time: 20+ hours per week, with flexible participation.
• Location: open to candidates in the United States.
• Language: fluent in English.
• Education: master's degree in physics.
• Skills: strong understanding of physics concepts and the ability to explain them clearly.
• Judgment to assess whether AI responses are accurate, relevant, and sufficiently reasoned.
• Ability to research topics and create clear, domain-specific prompts.
• Comfort working independently and primarily asynchronously.
• Relevant background may include doctoral study or candidacy together with valid U.S. work or training authorization.
• Examples of relevant authorization include F-1 or OPT, J-1, and H-1B status, subject to applicable program or visa requirements.
• New I-983 arrangements for STEM OPT are not currently accommodated. Already approved I-983 arrangements and pre-graduation OPT, CPT, J-1, and H-1B participation are identified as unaffected.

Apply on OpenTrain with your resume, then complete the application on the hiring site.

AI training is the human work behind modern AI systems, including writing prompts and reviewing model responses. OpenTrain is a platform where people find AI training and data-labeling projects, build a profile, and apply in minutes.