What the work actually involves

You'll be handed model outputs that look authoritative and asked whether they are right. In practice that means checking whether a stated boundary-layer transition criterion was applied at a Reynolds number where it holds, whether a normal-shock relation was used across an oblique shock, whether a claimed lift coefficient is physically achievable at the given angle of attack, or whether a described CFD setup (mesh resolution near the wall, y+ targets, turbulence model choice, boundary condition placement) would actually converge to something meaningful. Tasks vary by project: some batches are pure grading against a rubric, some ask you to write the corrected solution with worked steps, and some ask you to author new problems hard enough that current models fail them.

The hard part is rarely the physics. It is being explicit about why an answer is wrong — sign error, wrong regime assumption, plausible-sounding but fabricated correlation, right number by wrong route — in language a training pipeline can use. Vague feedback ("this is inaccurate") gets rejected in review.

What micro1 screens for

  • An AI-led interview first. micro1's screener conducts a live conversational interview, asks follow-ups based on what you just said, and will push on any claim you make. Depth is tested by follow-up, not by your CV.
  • Named, verifiable experience. Specific solvers (Fluent, STAR-CCM+, OpenFOAM, SU2), specific facilities, specific vehicle classes or aircraft programmes, specific published work.
  • Calibration under uncertainty. Whether you can say "this is defensible within ±15% and here is what would change that" rather than asserting or hedging.
  • Writing. Reference answers are read by other reviewers. Sloppy notation loses tasks.

Logistics

Fully remote and largely asynchronous, with tasks pulled from a queue and turnaround windows measured in days rather than hours. Most contributors run 10–20 hours a week alongside other work; some projects offer more volume in bursts. Rates in the $80–130/hr band have been observed and track specialisation, review quality, and whether you're grading or authoring — they are not guaranteed and are set per project. Expect a calibration batch before full task access.