What the work actually looks like

You receive AI-drafted legal content — a motion section, a case memo, an appellate argument, sometimes a response to a fact pattern with a deliberately buried ambiguity — and you assess it the way you would a first-year associate's draft that is about to go out under your name. That means checking whether the cited authority exists and stands for what the model says it stands for, whether the standard of review is correct, whether the argument survives the obvious counter, and whether the analysis holds up under the procedural posture it claims. Your output is usually a written critique plus, in many tasks, a corrected or model version of the passage.

A second stream of work is edge-case judgment: prompts where jurisdictions split, where the ethical answer diverges from the strategically attractive one, or where the correct response is that a competent lawyer would not answer without more facts. These are the tasks the project values most, because they are where models fail quietly.

What the platform screens for

  • Verifiable credentials. Bar number, jurisdiction, good standing, ABA-accredited JD. Clerkships, federal court admissions, and appellate practice get weighted.
  • Depth under follow-up. Screens ask about a matter you handled, then push on the procedural specifics — who bore the burden, what the standard was, why the other side's framing failed.
  • Error taxonomy. Can you distinguish a fabricated citation from a real case cited for the wrong proposition from a correct case applied under the wrong standard? Reviewers who write "this is wrong" without classifying the failure are not useful for model calibration.
  • Writing. Sample legal prose is frequently requested. Law journal editing, published scholarship, and teaching experience are treated as signal.

Logistics

Contractor, hourly, remote and US-only. Work is asynchronous with no fixed shifts, but projects run in waves and reviewers with reliable weekly capacity — often 10–20 hours — see steadier task flow. Expect iterative feedback from a cross-functional team and periodic rubric revisions you are expected to absorb. The $80–140/hr band reflects rates observed for this role; actual offers vary with credentials, project, and task type, and are not guaranteed.