What the work actually is

You pull a task from a pre-validated pool. Each one arrives pinned to a specific git commit, with a draft engineering question, positive and negative rubrics, foils, and a golden solution. Your job is to make that task hard in the right way: read the subsystem until you understand how it behaves at runtime, trace the PR and issue history that explains why it behaves that way, then rewrite the question so a frontier coding agent fails it while a competent senior engineer with repo access can still answer it fairly. You adjust rubrics and foils so that plausible-but-wrong reasoning is rejected on the merits, run the Check → Golden → Run → Analyze loop in Mercor Studio until every stage passes, document the task in a README, and respond to reviewer feedback on anything returned.

The repository set sits at the centre of the JS/TS ecosystem — Babel, Jest, Astro, TypeORM, Apollo Server, GraphQL-JS, Socket.IO, TanStack Router, NextAuth, XState, TypeDoc, typescript-eslint. The task mix leans toward architecture and system design (roughly 48%) and code onboarding (roughly 35%), so the questions that land are about how a subsystem is wired and why, not about what a type annotation says.

What the screen looks for

  • Real production TypeScript, not TypeScript-flavoured JavaScript: advanced type system behaviour, module resolution, bundler and transpiler pipelines, and the difference between compile-time and runtime semantics.
  • Evidence you have navigated an unfamiliar monorepo and come out able to explain runtime behaviour — plugin ordering, cache invalidation, transform pipelines, resolver execution.
  • Judgment about question difficulty: can you tell an agent-defeating question from an ambiguous one, and a good foil from a trick?
  • Written precision. The questions and rubrics are the deliverable; sloppy phrasing is a returned task.

Logistics

Fully remote, asynchronous, with expert team leads spanning US, India and Nigeria time zones. Work happens inside Mercor Studio; time is tracked with Insightful and access is provisioned through Okta. Onboarding requires identity verification, a background check, signed Terms of Work and CIIAA, a tax form, and a short calibration quiz before task access opens. Pay is per approved task — $130 as observed on this listing, not guaranteed — and approval runs through a second engineer plus a super reviewer. Volume is yours to set; there are no fixed hours, but tasks that come back for revision cost real time.