What the work actually involves

You'll spend most of your time in two modes. In the first, you write reference material: a lecture deck on a topic you teach, a conference presentation, a rubric, a course-design document — built the way you'd actually build it, with the reasoning behind slide order, chunking, and visual choices made explicit. In the second, you evaluate model output against a prompt someone else wrote, deciding whether a generated deck would survive contact with a real lecture hall. That means catching the twelve-bullet slide, the animation that reveals the answer before the question, the analogy that's fluent and wrong, the rubric whose criteria overlap so badly no two graders would agree.

Refinement sits between the two. You'll be asked to take a mediocre generated deck and rewrite it to a standard you'd be willing to present, then explain what changed and why — the explanation is often more valuable to the lab than the fix itself.

What the platform screens for

Ethos runs an AI voice screen before any human conversation. It verifies the concrete things first: terminal degree, institution, years teaching, courses taught. Then it probes depth — expect follow-ups that push past your first answer, asking why you sequenced a topic that way or how you'd know a slide failed. It also tests evaluation judgment specifically: distinguishing a deck that's wrong from one that's merely not how you would do it is the single skill that separates useful annotators from noisy ones.

  • Expect to name real courses and real material; vague gestures at "graduate teaching" score poorly
  • Expect to defend a pedagogical choice under pushback
  • Expect a direct question about weekly hours and how you'd log them

Logistics

Fully remote and asynchronous — no fixed meetings, no overlap requirement. Work is drawn down from a queue on your own schedule, typically 5–20 hours a week, with more available for people who want it. Pay is observed at $80/hour and is not guaranteed; rates vary by task type and project. Most contributors treat this as term-time-flexible work that scales up between semesters.