What the work involves

micro1 recruits domain experts to sit between a model and the people training it. For product management, that means writing and grading realistic artefacts: a discovery brief for an ambiguous request, a PRD with explicit non-goals, a prioritisation memo that names the trade-off it accepts, a metric tree that separates input metrics from vanity ones, or an escalation note to engineering when scope and date can no longer both hold. You will often be given two model responses and asked which is better and why — and the why is the deliverable. Rubric notes that say "more thorough" get rejected; notes that say "Response B invents a conversion baseline the prompt never supplied, and its success metric is measurable only after the launch decision is already made" are what the work pays for.

Expect a mix of task types across a project: writing gold-standard reference answers, red-teaming prompts where a plausible-sounding roadmap hides a bad assumption, and reviewing other contributors' annotations for consistency. Volume comes in waves tied to client projects rather than a steady weekly load.

What the screen looks for

  • Real operating history. Shipped products, named companies or product areas, decisions you owned rather than facilitated. Screeners follow up on specifics, and vague ownership claims collapse fast.
  • Judgment you can articulate. The ability to explain why a prioritisation call was right given what you knew at the time, and what would have changed it.
  • Comfort with rubrics. Consistency matters more than brilliance. Graders who drift between tasks cost projects more than graders who are merely strict.
  • Written precision. Nearly all output is written prose read by people calibrating a model. Hedged, padded writing is a disqualifier.

Logistics

Fully remote and asynchronous. Contributors typically commit 10–20 hours per week within a project window, scheduled around their own calendar, though some projects request overlap with US hours for calibration sessions. Engagement is contractor-based per project, not salaried. The pay band above reflects rates observed on this platform for senior product roles; actual rates vary by project, task type, and calibration performance, and are not guaranteed.

micro1's own posting for this role carried no written description, so the detail here reflects how comparable product-management evaluation projects on the platform are structured. Confirm scope and rate with the recruiter before committing time.