What the work involves

You review AI outputs against the kind of questions a BIM coordinator fields daily: whether a hard clash between a duct run and a structural beam is genuinely a conflict or an acceptable tolerance issue, whether a proposed sequencing change breaks a trade's install order, whether an LOD 300 model is being read as if it were LOD 400. Tasks typically arrive as prompt-and-response pairs to grade, side-by-side comparisons to rank, or blank prompts where you author the correct answer and the reasoning behind it. Some batches focus on software specifics — Revit families, Navisworks clash matrices, IFC export behavior, ACC/BIM 360 issue workflows — and others on judgment: RFI drafting, coordination meeting decisions, constructability flags a superintendent would actually raise.

Ratings need written justification. Reviewers who only mark answers wrong without explaining the mechanism or citing the standard usually don't stay on the roster.

What the platform screens for

  • Real project experience: which trades you coordinated, model sizes, delivery method (design-build, IPD, hard bid), and where you sat — GC, subcontractor, design firm, or owner's rep.
  • Depth under follow-up. Screens push past tool names into how you set clash tolerances, how you handled a federated model that wouldn't reconcile, and what you did when a coordinated model diverged from field conditions.
  • Ability to separate "the model says" from "the field allows" — the failure mode most AI answers exhibit.
  • Clear technical writing. Your justifications are training signal.

Logistics

Fully remote and asynchronous. Batches are released with deadlines rather than shifts, so most contributors fit 5–15 hours a week around project work; some weeks have no available tasks. Pay for this listing is undisclosed — micro1 sets rates per project and per reviewer, and observed rates in adjacent engineering review work have ranged widely. Confirm the rate, task volume, and payment cadence in writing before your first batch.