What the work actually involves

You are handed technical documents from your own domain — architecture proposals, CI/CD configurations, incident postmortems, API specifications, infrastructure-as-code walkthroughs, and increasingly model-generated versions of all of the above — and asked to judge them against a rubric someone else wrote. The output is not a thumbs-up. It is structured written feedback: what is wrong, where exactly, why an engineer would care, and what the corrected version should say. A good deal of the material arrives in Excel, Word, and PowerPoint, so comfort moving between those formats and producing clean, reviewable deliverables matters more than it sounds like it should.

Most tasks land in one of three shapes: rating a single artifact on several rubric dimensions with written justification, comparing two candidate outputs and defending a preference, or annotating a document line-by-line where the reasoning goes off the rails. Volume and task type shift between projects. You may spend a week on Kubernetes manifests and the next on distributed systems design reviews.

What the platform screens for

Mercor's screen is AI-led and follows up. It probes whether your engineering experience is load-bearing — expect to be asked what you actually built, at what scale, and what broke. Generic answers get one or two follow-up questions until they either resolve into specifics or don't. The other half of the screen is evaluation judgment: whether you can separate a document that is wrong from one that is merely written in a style you dislike, and whether you will apply a rubric you disagree with rather than substituting your own taste.

  • Depth in a named area: backend, infra, platform, SRE, developer tooling
  • Ability to articulate a defect precisely rather than gesturing at it
  • Consistency — will you score the same artifact the same way twice
  • Willingness to write, at length, in clear prose

Logistics

Fully remote and largely asynchronous, with tasks pulled from a queue against deadlines rather than fixed shifts. Contributors are typically eligible to work in the US, Europe, UK, or Canada. Many people run this alongside a full-time engineering job at 10–20 hours a week; some projects offer more. Pay is per hour of qualified work at rates observed in the $80–160 range — the band reflects what has been posted, not a guarantee for any given project.