What the work involves

When a project draws from this pool, the tasks tend to fall into three shapes. You write a teaching problem from material you have actually taught — specifying the grade or course level, the misconception it targets, the constraint that makes it non-trivial, and the reference answer a model's attempt will be scored against. You grade model-written explanations, worked examples, and assessment items against a rubric. And you flag the harder category: output that is factually correct but pedagogically wrong for the stated learner — an explanation pitched three reading levels too high, a worked example that quietly skips the step where students actually fail, an assessment item that measures recall while claiming to measure transfer.

Most of the deliverable is your written reasoning, not a score. A rating with no explanation is close to useless to a research team; a rating with a sentence naming the prior knowledge the explanation assumed and didn't check is the thing they're paying for.

What the screen looks for

The 20-minute AI interview and background verification are checking that you have taught the material yourself and can say, in specifics, where learners get stuck in it. Credentials and institution matter less here than whether you can name a misconception in your subject and describe how you diagnosed it. The screen also probes tolerance for ambiguity — these projects ship with imperfect instructions, and the platform wants people who raise the unclear case rather than guessing silently and producing inconsistent labels.

  • Concrete teaching history: subject, level, years, setting
  • Ability to explain a grading decision in writing, not just make one
  • Willingness to flag unclear rubric language

Logistics and honest expectations

This is a standing listing, not an opening. It does not accept or reject anyone — you will not receive a decision. Applying places you in the education and training pool; you get invited to a specific project listing, with its own named rate, hours, and client, when one matches your subject and level. That can happen within a week or take several months. The observed band across education projects on this platform runs roughly $30–85/hr and is set per project, not here. Work is remote and asynchronous, typically in deadline-bounded batches rather than fixed shifts, and sign-up is once across all Mercor networks you qualify for.