What the work actually is

You are handed advanced physics problems drawn from — or adjacent to — your specialization, and you produce the kind of solution you would put in an appendix of a paper: stated assumptions, explicit approximations, every non-obvious step shown, final results with units and limiting-case checks. Output is LaTeX. Where a derivation benefits from symbolic or numerical confirmation, you attach a clean SymPy or Jupyter workflow that a reviewer can re-run without guessing at your environment. A meaningful share of the job is not solving but reading: catching that a problem statement omits a boundary condition, silently assumes a regime it never names, or is dimensionally inconsistent — and then writing down a defensible interpretation rather than quietly picking one.

Solutions go to human reviewers, and revision cycles are normal. Cleanly integrating corrections — rather than re-litigating them or patching the text around them — is part of what gets contractors renewed.

What the screen is looking for

micro1's screening is AI-led and leans on verifiable specifics. Expect to name your subfield precisely, point to two to five arXiv or DOI records from roughly the last five years, and then be pushed on the physics inside them until the follow-ups reach the edge of what you actually did. Generalists who describe their field at survey level tend to stall; the platform is staffing named areas — high energy, mathematical physics, biophysics, statistical physics, condensed matter (moiré, magnetism, PXP/Rydberg), AMO and quantum optics, gravitation, cosmology, astrophysics, quantum information, optical properties of materials — and wants depth in one.

No AI or ML background is required or expected. The tooling bar is LaTeX fluency plus working comfort in Python/SymPy/Jupyter, and honesty about where your computational reach ends is treated as a signal rather than a deduction.

Logistics

  • Fully remote, worldwide, contractor engagement.
  • Roughly 10–15 hours a week across an 8–10 week project window; hours are asynchronous, including evenings and weekends.
  • Pay in the $100–200/hr band as observed on this listing — rates vary with subfield scarcity and task tier and are not guaranteed.
  • Expect an onboarding calibration task or two before volume ramps up.