Physical Scientist Talent Network

$60–80/hrrate as listed

Mercor maintains a standing pool of physicists, chemists, materials scientists, and other physical scientists who write and grade tasks used to train and evaluate frontier AI models. You apply once, pass an AI screening interview, and get matched to contract projects as labs request physical sciences coverage — typically 15–30 hours per week at an observed $60–80/hr.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work actually involves

Projects vary by client lab, but the core patterns repeat. You write problems drawn from real experimental or theoretical practice — a spectroscopy interpretation, a thermodynamic derivation, an error-propagation question on a noisy dataset — and supply a defensible reference solution with your reasoning shown. You then grade model attempts against that reference, flagging not just wrong answers but plausible-looking chains that use the wrong approximation regime, mishandle units, or invent a citation. Other assignments are comparative: two model responses side by side, and you explain which is better and why in language a non-specialist reviewer can audit.

The hardest part is usually rubric discipline rather than physics. Labs need consistency across dozens of annotators, so your written justification matters as much as your verdict. Expect feedback loops where a reviewer disputes your grade and you defend or revise it.

What the platform screens for

Verifiable depth. Mercor's AI interview probes your stated subfield with follow-ups. Generic answers about "research methods" surface quickly; specific instrumentation, sample prep, failure modes, and analysis pipelines do not.
Credentials on paper. A PhD or an MS with substantial industry or national-lab research experience is the practical floor. Publications, thesis work, and named techniques help matching.
Written clarity. Much of the deliverable is prose explaining a judgment. Screening weights how precisely you write, not just what you know.
Independence. Async work, little hand-holding, ambiguous instructions that you are expected to resolve and document.

Logistics

Fully remote and largely asynchronous, with occasional synchronous calibration calls depending on the client. This is an open talent-network application, not a specific opening — approval makes you eligible, and matching happens on a rolling basis as labs request physical sciences coverage, which can mean weeks between profile approval and a first invitation. Commitments observed in the 15–30 hr/week range; some projects run a few weeks, others recur for months. The $60–80/hr band is what has been observed on physical sciences work and is set per project, not guaranteed at application.

What they look for

PhD or MS in physics, chemistry, materials science, or a related physical science
Hands-on experimental research experience with instrumentation and data collection
Quantitative data analysis, uncertainty and error propagation
Scientific writing — clear, auditable explanations of technical reasoning
Independent async remote work with minimal supervision
Prior AI evaluation, rubric grading, or peer review experience (preferred)

Before you apply

Would you pass this screen?

Drawn from what this listing actually requires. Answered in your browser — nothing is sent to us, and we do not store your answers.

Do you have at least two years of professional or graduate-level hands-on experience conducting experimental research in a physical science discipline (physics, chemistry, materials science, astronomy, earth science, or equivalent)?
Physical science research experience · required
Can you write and execute code (in Python, R, MATLAB, or similar languages) to analyze experimental data, or have you used statistical software regularly for scientific data analysis in your research?
Quantitative analysis skills · required
Are you located in a timezone within +/- 12 hours of US Eastern Time and able to work 15–30 hours per week on a flexible, rolling-project basis?
Remote work capability · preferred
Are you authorized to work as an independent contractor in your country of residence?
Work authorization · required
Can you read, write, and speak English at a level sufficient to explain scientific concepts clearly in written reports and video calls with non-specialist audiences?
English communication · required

0 of 5 answered

Included with this listing

Interview prep — 10 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: Whether the candidate has hands-on experience running real experiments end-to-end, not just data analysis or theory—a core requirement for evaluating AI training in physical sciences.
A strong answer: The answer should name a specific study, describe the experimental design and method choices made, explain how data was collected and analyzed, and note where the work was published or presented. It must show independent decision-making about methodology.
Where people lose it: Candidates often describe only their role in a larger lab effort without clarity on what they personally designed or decided. Vague language like 'we ran tests' without specifying the apparatus, sample size, controls, or their own contribution suggests they may not have deep experimental ownership.