If you do science for a living — or you're a grad student or postdoc deep in a discipline — AI labs want your reasoning. Frontier models still make confident scientific errors, and the only way to catch and correct them is to have real physicists, chemists, and biologists evaluate the output. The work is remote, flexible, and pays $50–$200/hr depending on discipline and depth.
Here's the discipline-by-discipline picture, what's open, and how to land it.
What scientific AI-training work looks like
- Reasoning evaluation — judge whether the model correctly applied a physical law, reaction mechanism, or biological process to a problem.
- Problem & scenario design — write challenging questions (with correct solutions) that stress-test the model.
- Reference-answer generation — produce the gold-standard, publication-grade answer the model trains on.
- Red-teaming — surface where the model hallucinates citations, mangles units, or misapplies a theorem.
Discipline-specific roles open now
AfterQuery currently lists distinct science roles you apply to directly:
- Physics Expert — $50–$80/hr
- Chemistry Expert — $60–$80/hr
- Computational Materials Science Expert — $150–$200/hr
- Bioinformatics / Proteomics ML Expert — $150–$200/hr
- CRISPR & Gene Editing SME (PhD / Masters) — $50–$150/hr
- Climate / Atmospheric / Geologic Modeling — $150–$200/hr
Browse them on our AfterQuery listings page or filter by research / science.
The pattern: computational depth pays most
Notice the split above — "pure" discipline roles (physics, chemistry) sit at $50–$80/hr, while roles that combine the science with computation or ML (computational materials, bioinformatics, climate modeling) hit $150–$200/hr. If you have both the domain and the modeling skills, lead with the combination. The heavily computational roles overlap with our data scientists & ML engineers guide.
Which platforms to apply to
- AfterQuery — the deepest discipline-specific science catalog; apply to your exact field.
- Mercor — engine-matched frontier-lab work for senior/PhD scientists.
- Handshake AI — structured fellowships that fit grad-student and postdoc schedules (see the PhD guide).
How to position a science background
- State subfield + level. "Organic chemistry PhD candidate, synthesis focus" beats "chemist."
- List publications, methods, and tools. These are the verifiable signals labs weight.
- Flag any computational skills. They unlock the top-tier ML-adjacent rates.
Get started
No prior AI experience needed — see the no-experience guide — and the AI interview tactics are in our interview guide. Browse current science roles on the home page and read the complete AI training jobs guide for the full picture.