Advanced Physics Problem Solver (PhD/Postdoc)
Contribute deep physics expertise to train next-generation AI models by producing rigorous, LaTeX-formatted solutions and reproducible computational notebooks. Contract, remote role (~10 hrs/week for 8–10 weeks) paying $80–$140/hr for PhD-level researchers.
Generative Ai Rlhf
$80–$140/hr
Compensation
Worldwide
Eligibility
Entry
Experience
Jun 30, 2026
Posted
Open worldwide
About OpenTrain
OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. We connect skilled contributors with projects that shape how AI learns and behaves, helping you consolidate opportunities and build a durable portfolio of AI-training work.
Why AI training matters
AI training (data labeling and human feedback work) is the human side of building modern AI: people create, verify, and refine the examples models learn from. This project lets you directly influence model reasoning in a fast-growing field while working remotely and flexibly.
- Work is remote and flexible — fit tasks around other commitments.
- No prior AI experience required; domain expertise (physics) is the primary qualification.
The role
Client: OpenTrain is recruiting Physics Experts (PhD or advanced-stage PhD candidates) to provide high-quality, technical solutions that teach models to reason about advanced physics problems. This is a contract, part-time role focused on producing rigorous, reproducible solutions and computational verification.
Labeling scope includes RLHF and prompt/response writing (PROMPT_RESPONSE_WRITING_SFT) — your scientific solutions will be used as high-quality reference responses and training data for model instruction-following and evaluation.
- Employment type: Contractor, Part-time.
- Label types: RLHF, Prompt/Response writing (SFT).
What you'll do
Deliver precise, well-documented solutions to advanced physics problems from your research subfield. Emphasize clarity, reproducibility, and rigorous justification of each step.
- Write rigorous derivations and analyses using LaTeX for mathematical notation.
- Use SymPy, Python, and Jupyter to provide symbolic or numerical verification and clear computational notebooks where relevant.
- Document assumptions, approximations, boundary conditions, and dimensional checks for each solution.
- Identify ambiguities in problem statements and propose reasoned clarifications.
- Respond to reviewer feedback and iterate on submissions to incorporate corrections and refinements.
- Maintain professional standards in documentation and reproducibility consistent with research practice.
Requirements
You must meet every substantive requirement below; these are strict because the work will be used as authoritative scientific training data.
- PhD in physics or advanced-stage PhD candidacy with active research experience in a relevant subfield.
- Research expertise in one or more areas: High Energy Physics, Mathematical Physics, Biophysics, Statistical Physics, Condensed Matter (including moiré systems, magnetism, PXP/Rydberg), AMO/Quantum Optics, Gravitation, Cosmology, Astrophysics, Quantum Information, or Optical Properties of Materials.
- 2–5 recent representative publications (past ~5 years) with accessible arXiv links or DOIs.
- Proficiency with LaTeX for presenting mathematics and technical arguments.
- Proficiency with SymPy, Python, and Jupyter for symbolic/numerical verification and reproducible workflows.
- Demonstrated excellence in written technical communication and strong analytical skills.
Time, pay, and location
This is a remote role open worldwide. The project requests a commitment of about 10 hours per week across an 8–10 week period.
Pay is hourly, USD: range $80–$140 per hour (project data lists $140/hr typical; exact rate may vary by task or reviewer assessment).
- Typical weekly commitment: ~10 hrs/week.
- Project duration: 8–10 weeks.
- Location: Fully remote, worldwide.
- Payment: Hourly, USD $80–$140/hr.
How to apply
If you meet the requirements, prepare representative materials showing recent work and publications (arXiv/DOI links) and examples of LaTeX-written solutions or Jupyter notebooks. Applications will be reviewed for domain fit and demonstrated technical communication skill.
- Provide 2–5 representative publications (past ~5 years) with links.
- Include a short CV or profile noting PhD status, subfield, and relevant tools (LaTeX, SymPy, Python, Jupyter).
- Attach or link to 1–2 sample problem solutions or notebooks that showcase your approach to derivations and reproducible computation.