What the work involves

You write single-turn prompts in your own technical area and label each one benign, dual-use, or adversarial — the kind of distinction that separates a shotfirer asking about stemming length from someone fishing for a recipe in academic packaging. Prompts go to frontier models, responses come back, and you judge each one against a written policy standard: did the model answer a legitimate question fully, or hedge into uselessness? Did it refuse the genuinely dangerous request, or engage because the phrasing sounded like a literature review? Each task closes with a reference answer — what a correct response should have contained — plus the technical reasoning that justifies your call.

The failure mode Mercor is chasing runs in both directions. Models stonewall the routine questions a mining engineer, post-blast investigator, or range safety officer asks every week, then cooperate with requests that should have been shut down. Calibrating that boundary is domain judgment, not policy trivia, and it is why the panel is being built from people who have actually handled, characterised, or rendered safe the materials in question.

What the platform screens for

  • Two distinct profiles, both wanted: specialty materials chemists (pyrotechnic and propellant formulation, synthesis, performance testing) and engineers or operators with initiation, munitions, or propulsion system experience.
  • Verifiable operational history — hazardous device work, federal materials certification, forensic incident analysis, commercial blasting engineering, or effects engineering.
  • Written rationale quality. Every judgment needs an explanation a non-specialist reviewer can follow. Post-blast reports, published work, standards documents, or expert witness declarations are strong signals; bring a sample or link.
  • Clean information boundaries. The work must not draw on classified, export-controlled, NDA-bound, or prepublication-review material. Holding such obligations does not disqualify you — declare them and the scope gets adjusted.

Logistics

Fully remote and asynchronous, compensated per completed task rather than hourly, with observed rates in the $65–75 range depending on task type and complexity. Volume is batch-driven, so expect uneven weeks. You will be reading and writing about misuse scenarios in your field for sustained stretches; Mercor briefs experts on this beforehand and you can pause or step away at any point without penalty.