Mercor is assembling a panel of professional video editors to build post-production tasks used to train and evaluate AI agents. Each task gives an AI agent what you'd hand a junior editor: source footage, a brief, and a deliverable spec, plus a hidden answer key or rubric so the finished video can be scored on the result, not on a text answer.
\nYou will:
\nContribute original footage you shot and fully own (or source openly licensed footage) and author realistic editing briefs from it.
\nBuild each task end to end in your own editing software: cut, introduce defects or distractors where the task calls for them, and produce the reference edit a correct result is judged against.
\nWrite the grading criteria — concrete, checkable standards with the editorial reasoning behind them.
\nBlind-solve tasks built by other editors and write up a step-by-step solution.
\n5+ years editing professionally in film and TV, commercials, broadcast news, or creator and social production, owning deliverables end to end.
\nFluent in at least one professional editor (Premiere Pro, DaVinci Resolve, Final Cut Pro, Avid Media Composer), including basic color and audio cleanup.
\nComfortable with delivery specs: resolution, frame rate, codecs, audio channel layouts and frame-accurate cuts.
\nStrong plus: a library of unpublished original footage you shot and fully own, with people on camera who will sign an appearance release. The more, the better.
\nSubmit your resume and a link to your reel or portfolio.
\nComplete a short AI interview about your editing background.
\nAnswer a few questions about your footage: what you have, whether you own it outright, and whether the people in it consent to its use.
\nNot open to residents of California.
\nExpect about 12 hours/week.
\n