What the work involves

You are the expert judgement layer on top of a model that is trying to do design work. On a typical task you receive a product brief plus one or more model outputs — an onboarding flow, a settings hierarchy, a component spec, a redesign rationale, sometimes annotated screens or Figma-style descriptions — and you decide whether the output would survive contact with real users and real engineers. That means writing a critique that names the specific failure (touch target below 44px, modal used where a non-blocking inline state was called for, empty state with no recovery path, contrast ratio failing WCAG AA) rather than saying it "feels off."

Other task types on micro1's design queues include: ranking two competing model responses and defending the ranking, rewriting a weak model answer into the response an expert would have given, authoring fresh briefs and reference answers for training data, and reviewing another expert's critique for accuracy. Some projects ask for heuristic-evaluation-style scoring against a rubric the platform supplies; you are expected to apply that rubric rather than your own taste, and to flag where the rubric itself is ambiguous.

What the platform screens for

micro1's screening is AI-led and conversational — an interviewer that asks about your background, then follows up on whatever you claim. It rewards designers who can cite shipped products, describe the constraint they were designing under, and explain a decision they later reversed. It penalises portfolio narration and vocabulary without mechanism: if you say "we improved the information architecture," expect to be asked what the old structure was, what users failed to find, and how you knew. Accessibility literacy, design-system thinking, and the ability to separate a preference from a defect all come up.

Logistics

  • Fully remote and largely asynchronous; tasks are claimed from a queue rather than scheduled.
  • Most contributors commit 10–20 hours a week, though project loads fluctuate and dry spells happen.
  • Hourly, invoiced per reviewed task time; the $90–140 range reflects rates observed by contributors, not a guaranteed offer.
  • Expect a paid or unpaid calibration task after screening before live work opens up.