What the work involves

You receive legal or commercial documents — draft agreements, clause libraries, compliance memos, diligence summaries, term sheet comparisons — alongside model-generated analysis of them. Your job is to say, in writing, where the model is right, where it is wrong, and why a practitioner would notice. That means catching a misread indemnity cap, a governing-law clause applied to the wrong entity, a redline that changes allocation of risk without flagging it, or a summary that is fluent and confidently incorrect. Some tasks ask you to write the reference answer yourself; others ask you to score two candidate outputs against a rubric and justify the preference in a few tight paragraphs.

Output is heavily document- and spreadsheet-based. Expect to work in Word with tracked changes, log ratings and error categories in Excel, and occasionally build a short deck summarising failure patterns across a batch. The rubric is not yours to redesign — a large part of the job is applying someone else's taxonomy consistently across dozens of items, including the ones that fall awkwardly between categories.

What the platform screens for

Mercor's screen is AI-led and interview-style. It probes whether your contract experience is real and specific: what you actually negotiated, which clauses you owned, what jurisdiction and deal size, what went wrong. It then tests evaluation judgment — whether you can separate "this is not how I would draft it" from "this is legally wrong," and whether you can grade against a rubric rather than your own house style. Vague seniority claims do poorly; a concrete story about a limitation-of-liability negotiation does well. Expect follow-ups that push one layer deeper than your first answer.

Logistics

  • Fully remote and largely asynchronous; work is claimed from a queue rather than scheduled.
  • Availability is usually stated in hours per week — 10–20 is common, and consistency matters more than volume.
  • Eligibility is limited to the United States, Europe, the UK, or Canada.
  • Pay is as observed on the platform and varies by project, task type, and assessed expertise; nothing here is a guaranteed rate.