The work

Project Rachel publishes legal-benchmark tasks and collects commentary on them from outside readers. Your job starts after publication: someone has argued that a prompt is ambiguous, that a rubric line is wrong on the law, or that the attached documents don't support the graded answer. You decide whether the objection holds, write a short reasoned disposition, and then run structural QC across the whole item — does the prompt state the governing regime and the posture, does each rubric criterion map to something actually determinable from the source set, are the exhibits complete and correctly cited.

Subject matter runs across the practice area rather than one corner of it: data processing agreements and controller/processor allocation, consent mechanics and notice design, cross-border transfer analysis (SCCs, transfer impact assessments, adequacy), and incident response including notification triggers and timelines. Source bundles are large — 10 to 100 documents per task — and much of the QC is confirming that the rubric's claimed answer survives contact with all of them, not just the two that were obviously drafted for it.

What the platform screens for

  • Whether you can say what the law actually requires under a named regime, unhedged, and then defend it when the screener pushes back with a plausible counter-reading.
  • Whether you distinguish a genuine defect in a rubric from a commenter who simply prefers a different answer — overturning too readily is as much a failure as rubber-stamping.
  • Whether you can work at document-set scale without losing the thread: locating the one clause in a 60-document bundle that changes the analysis.
  • Practising or recently-practising depth. Generalist privacy familiarity reads thin under follow-up here.

Logistics

Fully remote and asynchronous. Reviewers generally pick up task batches on their own schedule, with turnaround windows rather than fixed hours; sustained availability of roughly 10–20 hours a week is the pattern that fits the workload. The $120/hr band is what has been observed on this project — rates on Mercor vary by task type, seniority, and review tier, and nothing here is guaranteed. This is a private shortlist listing: candidates are added directly by the project team and it is not open for general application.