What the work involves

You build the test cases frontier labs use to find out whether a model actually understands care coordination. That means authoring scenarios that run from admission through discharge and post-acute placement — a 78-year-old with a hip fracture, a Medicare Advantage plan, a two-story house and no caregiver at home; a self-pay patient needing IV antibiotics; a behavioral health hold with no accepting facility in county. Each scenario comes with a reference answer and, more importantly, the reasoning: why SNF and not home health, which criteria set supports it, what the payer will and won't authorize.

The second half of the job is grading. You read model-generated discharge plans and level-of-care determinations and score them for clinical appropriateness and benefit accuracy. The failures that matter most are the plausible ones — the plan that reads like competent documentation but sends the patient somewhere their benefit doesn't cover, ignores a custodial-care exclusion, or would draw a denial on the first review. Writing up why it fails, in language an engineer with no clinical background can act on, is the deliverable.

What the platform screens for

  • A verifiable active credential. Active RN or LCSW license, or CCM certification. This is checked.
  • Real case management reps. Two or more years in a hospital, payer, or community setting — discharge planning, utilization review, or complex care coordination specifically, not general floor nursing or clinical social work.
  • Criteria fluency. InterQual or MCG experience, payer-side UM or appeals work, and CCM/ACM certification are preferred and tend to raise the band.
  • Writing under scrutiny. Screens include follow-ups on your own answers. Vague reasoning that can't survive a "why not the other level of care?" probe is the most common failure point.

Logistics

Fully remote, fully async — no caseload, no on-call, no scheduled shifts. You set your hours week to week with a floor of 10. Pay is hourly via Stripe on a weekly cycle; the $70–115 range reflects rates observed on the platform and varies with certification, criteria experience, and task type. Work is ongoing project-based rather than a fixed contract term.