What the work actually involves

You will spend most of your time reading AI output and deciding whether an experienced IT manager would sign off on it. That means grading a generated incident response for whether the escalation path, comms cadence, and rollback criteria hold up under a real Sev-1; checking whether a project plan accounts for change freezes, dependency chains, and the fact that the DBA team is already committed elsewhere; and judging whether a technical standard is enforceable or just aspirational prose. A large share of tasks are head-to-head comparisons: two responses to the same scenario, and you pick the better one and write the reasoning. The written justification is the deliverable — a preference with a thin rationale is worth little to the model trainers, and reviewers read for whether your critique names a specific operational failure rather than a stylistic one.

Scenarios lean toward the coordination side of the job: system prioritization when two business units both claim urgency, vendor management and SLA interpretation, tier-2/tier-3 support routing, team assignment under headcount constraints, and budget or lifecycle decisions. Depth matters more than breadth — you are not expected to know every stack, but you are expected to catch when an AI confidently proposes something that would violate change management, ignore a compliance obligation, or quietly assume staff who don't exist.

What the platform screens for

Mercor's screen is AI-conducted and follow-up heavy. It will take a claim from your resume — "managed a 12-person infrastructure team," "owned the ITSM rollout" — and press on the specifics: headcount, reporting lines, budget authority, what broke and what you changed. Vague seniority language collapses fast under that questioning. Expect at least one live scenario where you must critique a plausible-sounding but flawed IT artifact and say precisely why it fails. Availability is verified too: 20+ hours per week is a real threshold, not a preference, and the first-task timeline gates the move from task-based to hourly pay.

Logistics

  • Fully remote and largely asynchronous; you pick up tasks from a queue rather than attending meetings.
  • Immediate start, roughly 4–6 weeks in duration, with possible extension depending on project volume.
  • Payment begins task-based on first task approval; completing that first task inside the required window qualifies you for hourly pay for the remainder.
  • Pay in the $70–110/hr range is what has been observed on comparable Mercor management-domain projects, not a guarantee — actual rate depends on the screen outcome and project tier.