What the work actually involves
Projects vary, but they tend to fall into three shapes. The first is authoring: you take a matter you handled, strip it to facts, forum, and governing law, write the question a model should be able to answer, and then write the reference answer it gets graded against — including why the wrong answers are wrong. The second is grading: a model has drafted an indemnity provision, made a privilege call, or cited authority for a proposition, and you decide whether it holds. The third is the harder one — identifying where a model states black-letter law accurately and would still lose in front of this judge, under this standard of review, on these facts.
What the screen is looking for
Mercor's AI interview runs roughly 20 minutes and is built to verify that your experience is real and specific. Expect it to ask what you actually did on matters, not what your firm did. The platform screens for professional experience in practice, compliance, regulatory work, or policy; for written reasoning that survives follow-up questions; and for comfort working from ambiguous instructions without either freezing or silently guessing. Flagging a bad prompt is treated as a signal of quality, not as a complaint.
Scope of the pool
- Litigation and criminal justice; general practice
- Corporate and M&A; intellectual property
- Compliance, risk, and regulatory affairs
- Policy work, in-house and government
- Paralegals and legal assistants — this pool is deliberately broader than practising attorneys
Logistics and honest expectations
This is a standing listing, not a job opening. You apply once, confirm where you work from, complete the interview, and sit in the pool. There is no accept or reject decision on this listing — you hear from Mercor when a project matches your area, which may be next week or several months out. Actual rate, hours, and client are named on the project listing that follows, and any hiring decision is made there. Rates across the legal field have posted between $60 and $150 an hour, set per project by scope and depth; that is an observed range, not a guarantee. Work is remote and largely asynchronous, usually in blocks of a few hours against a deadline rather than fixed shifts.