What the work involves

You are asked to judge whether a model's mechanical reasoning would survive contact with a real shop floor. Typical tasks include grading two competing AI responses to a question about press-fit tolerances, GD&T callouts, or bearing failure modes; rewriting an answer that cites the wrong material spec or an unsafe lockout sequence; and writing original prompts drawn from jobs you have actually run — a misaligned coupling, a fixture that chattered, a pump that cavitated after a rebuild. Some batches focus on reading drawings and BOMs, others on troubleshooting narratives, others on documenting procedures step by step. You are expected to explain why an answer is wrong in terms a reviewer can verify, not just mark it down.

What the platform screens for

micro1's screening is AI-led and conversational. It probes whether your experience is specific and recent: which machines you ran, what tolerances you held, which standards you worked to (ASME Y14.5, ISO 2768, OSHA lockout/tagout), and what you did when a part came back out of spec. Expect follow-up questions that press on details a generalist would fumble — surface finish units, thread class fits, torque sequence logic. Credentials help, but demonstrated hands-on judgment matters more than a résumé line.

Logistics

  • Fully remote and asynchronous; you claim tasks from queues rather than working fixed shifts.
  • Volume fluctuates by project — some weeks offer 5–10 hours, others 20+. Treat it as supplemental, not a salary replacement.
  • Pay is hourly within an observed $30–70/hr band; the actual rate depends on project, specialization, and calibration performance, and is not guaranteed.
  • Reliable internet, comfort typing detailed written rationales, and a willingness to redo work after calibration feedback are practical requirements.