Mercor keeps a standing pool of materials scientists who write and grade problems used to train and evaluate frontier AI models. Work is project-based and intermittent: you author problems from systems you actually processed and characterized, then judge whether a model's structure–property reasoning and material selections would survive service conditions.

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

What the work involves

Projects fall into three rough shapes. You author a materials problem drawn from work you personally did — the alloy or polymer or ceramic system, the full processing history, the property target — and you write the reference answer that a model's attempt will be graded against. You grade model-written structure–property reasoning, asking whether the proposed microstructure actually explains the measured property, whether the characterization cited supports the claim being made, and whether the processing window is one a real furnace, press, or reactor could hold. And you flag the failure mode reviewers without hands-on experience miss: the material that meets spec on a datasheet and dies in service to fatigue, corrosion, thermal cycling, or the simple fact that nobody can manufacture it at volume.

What the platform screens for

The AI interview runs roughly 20 minutes and is looking for evidence you handled the material yourself rather than read about it. Expect follow-ups that push for specifics: which heat treatment, which characterization technique and why that one, what the data actually showed versus what you expected. A graduate degree in materials science or an adjacent field is the stated baseline, paired with research or industrial experience. Because nearly all deliverables are written, the screen also weighs whether you can explain a judgment call in prose that a non-specialist reviewer can audit — and whether you will say "this task is underspecified" instead of guessing.

Logistics and the standing-pool model

This is not a job opening. It is one permanently open listing per field, from which Mercor draws when a client project starts on short notice. You will not receive an accept or reject decision here. After applying, confirming your work location, and completing the interview, your background is verified and you enter the materials science pool; invitations to specific project listings — each naming its own rate, hours, and client — may arrive within a week or after several months.

Fully remote and asynchronous, with hours set per project
Observed past rates: $60–$120/hr, set per project by scope and depth, not guaranteed
Completing the domain expert interview and joining every network you qualify for both improve match odds; one signup covers all of them

Materials Science Expert

What the work involves

What the platform screens for

Logistics and the standing-pool model

Interview prep — 10 questions