What the work involves
You will be handed construction problems — a set of drawings and specs, an RFI, a sequencing conflict, a submittal that does not match the detail — and asked to work them the way you would on a live job, but with your reasoning written down step by step. Some tasks ask you to produce the problem and the defensible answer; others ask you to grade or correct a model's attempt at one. Typical outputs include a documented review of a drawing package, a written resolution to a field conflict with the code or standard citation that supports it, or a schedule-logic critique explaining why a proposed sequence breaks.
The distinguishing requirement is reproducibility. A correct conclusion with no traceable path from the contract documents to that conclusion is not useful here. Reviewers want to see which sheet, which spec section, which code provision, and which assumption you made when the documents were silent — because the whole point is teaching a model where a judgment comes from.
What the platform screens for
- Whether you have actually held a construction engineer, project engineer, or field engineer seat, and can describe specific projects: type, scale, your scope, what you owned.
- Depth under follow-up on fundamentals — sequencing, means and methods, materials, tolerances, and the code and standard families (IBC, ACI, AISC, ASCE, OSHA) you claim.
- Evaluation judgment: can you tell a wrong answer from a right-looking one, and articulate precisely why.
- Honest availability. The window is short and the throughput expectation is real.
Logistics
Fully remote and asynchronous — no standing meetings, no fixed hours. Projects are offered individually and you accept or decline; accepting implies 10–20 hours that week. The engagement as described runs 2–3 weeks, with further work typically contingent on quality scores from your first batch. Pay is stated as $50–100/hr as observed on the platform, not guaranteed; rates vary by task complexity and reviewer rating.