What the work actually involves
This is evaluation work, not production design. On a typical shift you are given a design brief plus one or more model attempts at it — a dashboard layout, a mobile checkout flow, a component variant set, a landing page in HTML/CSS or Figma-like structure — and you rank the outputs, then explain the ranking in writing that a model can learn from. "The hierarchy is wrong" earns nothing; "the primary CTA and the secondary link share the same weight because both use 16px semibold at the same contrast, so the eye has no entry point" is the unit of value. Other task types include writing new design prompts with a defensible gold-standard answer, flagging accessibility failures (contrast ratios, touch target size, focus states), and reviewing whether generated code matches the intended visual spec.
What the platform screens for
micro1's intake runs as an asynchronous AI interview that transcribes your answers and probes follow-ups, so vague claims get tested rather than accepted. Expect it to push on:
- Portfolio specifics — which screens you personally designed, what the constraint was, what you changed and why
- Whether you can articulate visual judgment in causal terms rather than taste language
- Working fluency with design systems, tokens, and how design decisions survive handoff to engineering
- Accessibility knowledge at the level of applied standards, not slogans
- Whether you can hold a consistent standard across dozens of near-identical outputs without drifting
Logistics
Fully remote and largely asynchronous, with task batches claimed from a queue. Most contributors run 10–20 hours a week; some projects offer more during ramp-up and then taper. Calibration sessions and rubric updates are occasional and usually recorded. Reviewed output volume, not hours logged, tends to drive continued allocation, and rates are set per project — the band reflects what has been observed rather than a guarantee.