What the work actually is

This is a research study, not annotation or model grading. Through Project Mercator, Mercor is assembling a detailed map of what public-defense work consists of at the task level. In the first survey you describe your primary responsibilities, then decompose them: the discrete tasks inside each responsibility, the smaller sub-steps inside those, and — importantly — which tasks can't start until another task produces an output. Arraignment coverage, discovery review, client interviews at the jail, plea negotiation, suppression motion drafting, trial prep, sentencing mitigation: the study wants the version you personally performed, in the jurisdiction and period you practiced, including variant forms of the same task (a misdemeanor plea conference is not a homicide plea conference).

Written instructions and a study coordinator are provided. No outside research is expected, and no real work product is required or wanted — do not paste anything from a real file. Client names, case numbers, identifying facts, and privileged material are all out of bounds; the study is designed so that none of it is necessary.

What the screen looks for

  • Genuine public-defense practice. Hands-on representation of indigent clients in the United States. Court-appointed and contract public-defense arrangements count, but you'll be asked to explain the arrangement. Prosecution-only, law-school clinic-only, legal-content annotation, or general legal practice without public-defense experience won't qualify.
  • Specificity about your own docket. Jurisdiction, court level, offense mix, rough caseload volume, and years practiced. Vague gestures at "criminal defense work" read as thin.
  • Ability to decompose work. The core skill here is describing a workflow in ordered, dependency-aware steps rather than in narrative. Candidates who can only tell war stories struggle with the survey format.
  • First-person, non-AI answers. Applications and survey responses are expected to be written from your own experience.

Logistics

Fully remote and asynchronous. Wave 1 is capped at 90 minutes of reviewed survey time at $100/hr — $150 maximum — with reading and thinking time counted while the survey is open and focused; you can pause for breaks, and you contact the coordinator before working past the cap. Applying does not enroll you: applications are reviewed before personal survey links go out. A later wave may ask selected participants to review a merged task tree and add detail, with scope and pay confirmed before that work begins. Pay is as observed on the platform, not guaranteed.