What the work involves

You build the test material that measures whether an AI model can reason like a senior mechanical engineer inside a large industrial organization. In practice that means authoring realistic scenarios — a design review with conflicting stakeholder constraints, a thermal margin problem on a production program, a DFM escalation late in a tooling cycle, a failure investigation with incomplete field data — and then writing the reference output a strong engineer would produce. Each task is paired with a rubric that spells out what separates defensible engineering judgment from textbook recall: correct assumption-setting, appropriate safety factors, awareness of tolerance stack-up, knowing when a hand calculation is sufficient and when FEA or CFD is warranted.

Tasks span product design and CAD workflows, thermal and fluid systems, structural and stress analysis, manufacturing process engineering, and reliability and failure analysis. Scenarios often reference tools you would have used in an enterprise stack — SolidWorks, CATIA, ANSYS, MATLAB/Simulink, Teamcenter, Windchill — and methodologies such as GD&T, DFM/DFA, DFSS, and certification pathways under ISO, FAA, or FDA regimes. You may also review or calibrate against other contributors' work.

What the platform screens for

Mercor's screening is AI-led and leans heavily on specificity. Expect to be asked what you personally owned rather than what your team shipped: which program, which subsystem, what the constraint was, what you decided, what it cost. Follow-up questions probe depth — if you cite an FEA result, be ready to defend mesh strategy, boundary conditions, and validation against test. The strongest signal is a candidate who can explain why a technically correct answer would still be rejected in a real design review, since that is exactly the judgment the rubrics need to encode.

Logistics

  • Fully remote and asynchronous; no fixed shifts.
  • Contract work, typically 10–20 hours per week, with volume varying by project.
  • Observed pay of $70–80/hr — a band reported for this listing, not a guarantee; final rates depend on screening outcome and project.
  • Writing volume is substantial: expect to produce detailed technical prose, not just numeric answers.