What the work involves

Once matched to a project, your day-to-day is closer to writing hard exam questions than to bench work. Typical deliverables include authoring biology problems that a strong model gets wrong, building rubrics that separate a defensible answer from a plausible-sounding one, grading model outputs against those rubrics, and writing rationales that explain why a response fails — a mislabeled control, a conflated mechanism, a statistical test that doesn't fit the design. Some projects lean toward multi-turn conversations where you probe a model on protocol troubleshooting or data interpretation; others are batch review of existing responses. Subject matter tracks the raw posting: molecular and cellular biology techniques, experimental design, and life-sciences data analysis.

What the platform screens for

Mercor's screen is an AI-led video interview plus resume verification. It rewards specificity: named techniques you've personally run, the organism or system you worked in, what went wrong and how you diagnosed it. Vague claims get probed with follow-ups, and answers that stay general tend to end the conversation early. Reviewers are also assessing writing clarity — much of the deliverable is prose — and whether you can hold a defensible line on ambiguous questions instead of deferring to whatever the model said.

Logistics and pay

  • Fully remote and largely asynchronous, with deadlines rather than fixed shifts.
  • Typical commitments run 15–30 hours per week when you're on a project; the network itself carries no guaranteed volume.
  • Rates observed in this band run $60–80/hr, varying by project, seniority, and client. Not guaranteed — the actual rate is set per engagement.
  • Contract work, invoiced through Mercor. You may sit in the network for weeks before a matching project opens.

This suits practicing or recently practicing biologists — PhD candidates, postdocs, industry scientists — who want variable-hour contract work alongside existing commitments, and who are comfortable being evaluated on written reasoning rather than credentials alone.