What the work actually looks like

You are not delivering a training program to human learners. You are producing the artifacts an expert instructional designer would produce, at a quality level a model can learn from — and, often, judging whether a model's attempt at the same artifact is any good. A typical task might be: given a short brief about an audience (new retail supervisors, contact-centre agents, clinical onboarding), write measurable learning objectives, map a module pathway from those objectives through activities to assessment, then draft a rubric that would actually discriminate between a weak and a strong learner response. Other tasks flip it: two model-generated lesson plans, and you explain which is better and precisely why — vague verbs, objectives that can't be assessed, activities that don't serve the stated outcome, rubric levels that collapse into each other.

  • Writing objectives that are observable and measurable rather than aspirational ("understand" vs. "classify", "be aware of" vs. "produce")
  • Building alignment chains: objective → activity → assessment item, with the link made explicit
  • Drafting quizzes, scenario items and analytic rubrics, including distractor rationale
  • Adapting content for asynchronous digital delivery rather than facilitator-led rooms
  • Written critique of AI output against a rubric, with reasoning legible to a reviewer who wasn't there

What the screen is looking for

micro1's process is AI-led: an asynchronous conversational interview, follow-up probes on whatever you claim, and usually a short written or task-based sample. The screen is trying to separate people who have designed and shipped learning programs from people who can recite Bloom's taxonomy and the ADDIE acronym. Expect to be asked for a specific program you built, who the audience was, what you measured, and what you changed after the first cohort. Frameworks named without a concrete instance attached read as filler. Written quality matters more than usual here, because written output is the deliverable — long, hedged, jargon-dense prose scores badly even when the underlying thinking is sound.

Logistics

Remote, contractor, hourly at an observed $40–75/hr depending on assessed depth and the customer's task tier — not a guaranteed rate, and not guaranteed volume. Work is largely asynchronous and self-scheduled, with occasional live syncs or calibration sessions in US-overlapping hours. Volume on these projects moves in waves: a heavy week during a labelling push, quiet stretches between customer batches. Most contributors treat it as part-time alongside other work, and the people who stay on projects are the ones who answer reviewer feedback and re-calibrate quickly rather than defending their first draft.