What the work involves
You write engineering problems the way they actually arrive in practice — a requirements statement, a load profile, a set of constraints, an ambiguity someone has to resolve — and then you write the reference answer, including the reasoning that connects each decision to the requirement or the standard behind it. A scenario might run from a service entrance sizing calculation through protective device coordination, or from a sensor front-end spec through component selection, tolerance stack, and a design review that catches what the first pass missed.
The second half of the work is grading. You read model output and judge whether it is technically correct, whether it cites the right clause or applies the right derating factor, and — the part labs care most about — whether it is the kind of answer that reads fluently and would still fail on the bench or fail inspection. Wrong-but-confident output is the signal; your job is to name precisely what breaks and why.
What the platform screens for
AfterQuery's screen is AI-led and leans on follow-up questions. Expect to be asked what you personally designed, at what voltage or scale, which standard or code edition governed it, and what you would check first if a number looked off. Depth in one area beats a survey of many. Screeners also probe calibration: whether you can distinguish a real error from a defensible alternative approach, and whether you would flag ambiguity in your own scenario rather than grade a model against an underspecified prompt.
Logistics
- 100% remote, fully async — no standing meetings, no fixed hours
- Minimum 10 hours per week; you set your own schedule around it
- Weekly payment via Stripe
- Ongoing project work, with scenario batches assigned as labs request coverage
- Written output is the deliverable, so clear English documentation matters as much as the engineering