What the work involves

You receive artifacts an AI model produced in response to a realistic professional brief — a wetland delineation summary, a species inventory workbook, an EIA chapter, a client-facing slide deck on carbon accounting or habitat restoration. Your job is to read them the way a senior reviewer would: check whether the taxonomy, units, sampling logic, regulatory citations, and statistical claims hold up; flag fabricated references and quietly wrong numbers; and judge whether the deck or spreadsheet would actually survive being sent to a regulator or a client. Scoring happens against a rubric, but the value you add is the written justification — specific, structured, and traceable to a defect rather than a general impression.

What the platform screens for

  • Verifiable depth. Expect follow-ups on your stated specialty — methods you've run, standards you work under, what you'd expect a field dataset to look like. Vague answers stall the screen.
  • Evaluation judgment. Can you distinguish a plausible-sounding error from a genuine one, and rank severity when a document has both a citation problem and a formatting problem?
  • Artifact literacy. Slides and spreadsheets matter here. Reviewers who can only assess prose tend to miss chart mislabeling, broken formulas, and figure-caption mismatches.
  • Writing. Feedback must be legible to a non-specialist reader without losing technical precision.

Logistics

Fully remote and asynchronous. Work arrives in batches, so volume fluctuates — some weeks offer 10–20 hours, others little. Most contributors treat this as a supplement to existing practice or research work rather than a primary load. You'll need reliable access to Google Workspace and Microsoft Office, and comfort working inside a review tool alongside the artifact. Pay is hourly against logged review time; the $80–120/hr band reflects observed rates and is not guaranteed for any individual engagement.