What the work actually involves

You are not reviewing AI output all day; you are building the exam. A typical task starts with a plausible project scenario — a retaining wall on marginal soils, an intersection sight-distance check, a detention basin sized against a local drainage manual, a foundation bearing check off a geotechnical log — and ends with a complete, self-contained package: the prompt, the supporting artifacts, the defensible correct answer, and a multi-item rubric an LLM grader can apply consistently. The artifacts matter as much as the question. You will source or construct drawings, spec sections, geotech reports, cost estimates, CSVs and spreadsheets that look and behave like real project documentation, including the messiness (unit inconsistencies, a detail that contradicts the spec, data spread across three files) that makes a problem genuinely hard.

The hardest part is usually the rubric. A task is only useful if two different graders reach the same verdict, so each rubric item must be a discrete, checkable claim — the correct code edition cited, the load combination applied, the governing constraint identified, the check performed against the right allowable — rather than a holistic impression of quality. Expect iteration with project stakeholders on clarity, realism, and whether your "one right answer" survives scrutiny when an engineer with different regional practice reads it.

What the screen looks for

  • Breadth plus a real specialty. The scope spans structural, transportation, water, and site work. You are not expected to own all four, but you should be able to name which you can author to a licensable standard and which you would decline.
  • Code fluency you can cite. Not just "I use ASCE 7" but which edition, which chapter, and what changed that affects an answer.
  • Documentation depth. Evidence you have prepared, stamped, reviewed, or defended real technical submittals — plan sets, calculation packages, spec sections, engineer's estimates.
  • Unambiguity instinct. Whether you can spot the assumption that makes a problem have two defensible answers, and close it.
  • Written clarity. Everything you deliver is prose and tables that someone else must apply without asking you questions.

Logistics

Fully remote and asynchronous — no fixed hours, but there is a stated weekly minimum number of accepted submissions, so this is not a role you can dip into once a month. Pay is per task that meets specification, and the observed band for this listing is $50–100/hr equivalent; effective rate depends heavily on how fast you build artifacts and how often tasks pass review on first submission. micro1 says it fills these roles within about 48 hours and expects first tasks within 24–48 hours of onboarding, so availability in the immediate term is part of the assessment. A PE licence is strongly preferred rather than required; active pursuit of licensure is treated as credible.