Architects author design and documentation scenarios — schematic design through CDs — that frontier AI labs use to test model judgment, then grade model output for design quality and code compliance. Work is remote and fully async at a minimum of 10 hours per week, with observed pay of $100–125/hr paid weekly.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work involves

You write architecture problems the way they actually arrive in practice: a site with a setback conflict, a program that doesn't fit the egress geometry, a wall assembly that needs to hit an hourly rating and a U-value at once. Each scenario comes with a reference answer that spells out the reasoning — which code section governs, why one detail was chosen over another, what a plan reviewer would red-line. Then you grade model responses against that reference, and the harder half of the job is the last bullet in the listing: catching output that reads like competent architecture but would fail plan review, misapply an occupancy classification, or specify an assembly that cannot be built as drawn.

Most tasks land somewhere in schematic design, design development, or construction documentation. Expect a mix of authoring and review, with feedback loops from lab-side reviewers on whether your rubrics discriminate between a good answer and a merely fluent one.

What the platform screens for

Verifiable practice history. Project types, your role on them, phases you personally carried. Vague seniority claims don't survive follow-up.
Code fluency under questioning. Not memorized section numbers, but the ability to reason from occupancy and construction type toward a defensible answer, and to say which code and edition you work in.
Grading judgment. Can you separate a design choice you'd have made differently from one that is actually non-compliant? Reviewers who mark stylistic disagreement as error are expensive.
Written clarity. The reference answer is the deliverable. If the reasoning isn't legible to a non-architect reviewer, it isn't usable.

Logistics

Fully remote, fully async — no standing calls, no client hours. You set weekly hours above a 10-hour floor; pay is weekly via Stripe. Work is ongoing rather than a fixed engagement, and volume typically depends on which task families a lab is prioritizing that cycle. Revit or BIM depth and multi-building-type experience are preferred, not gating.

Architect Expert

What the work involves

What the platform screens for

Logistics

Interview prep — 10 questions