What the work involves

Bilingual language projects on micro1 typically ask you to judge machine-generated Dutch against a source or a prompt and then fix it. On a given shift you might rank two Dutch responses to the same instruction, rewrite a translation that is grammatically clean but reads like it was written by a machine, flag where a model has silently dropped a clause, or write reference Dutch answers that later become the standard other outputs are scored against. Guidelines are detailed and change between projects — a task that rewarded literal fidelity last week may reward natural register this week, and reviewers care that you follow the current rubric rather than your own preference.

The hard part is rarely spotting outright errors. It is deciding whether a plausible-sounding Dutch sentence is actually wrong: false friends carried over from English, register drift between u and je, Belgian versus Netherlands usage, compound spelling, or a translation that is defensible in isolation but wrong for the document's audience. Written justification matters — you are usually expected to explain in a sentence or two why an output was downgraded, in terms a non-Dutch-speaking project lead can act on.

What the platform screens for

  • Genuine near-native Dutch plus strong English comprehension, since source material and guidelines are almost always English.
  • Evidence rather than assertion: where you learned Dutch, where you've used it professionally, whether you write it as well as speak it.
  • Rubric discipline — whether you can apply someone else's error taxonomy consistently instead of freestyling corrections.
  • Calibration under follow-up: screeners commonly re-ask a judgement to see whether you hold a defensible position or fold.

Logistics

Fully remote and largely asynchronous. Work arrives in batches, so volume is uneven — some weeks offer steady hours, others little. Most contributors are asked for a realistic weekly commitment (often 10–20+ hours) and reliable turnaround inside stated deadlines rather than fixed shifts. A laptop, stable connection, and comfort with browser-based annotation tools are assumed. Rates are hourly and vary by project complexity; the band shown reflects observed postings, not a guarantee.