What the work actually involves
You spend most of your time reading: outpatient encounter notes, AI-drafted SOAP notes, visit summaries, and structured EHR extracts. For each item you judge whether the documentation reflects what a practicing PCP would actually write — correct assessment and plan, no invented findings, no dropped medications or allergies, appropriate follow-up. Then you write it down in a form an engineer can use: which span is wrong, why it is clinically wrong, and what the correct documentation would say. A meaningful share of the work is edge cases — a note where the AI hedges on a differential, or summarizes a patient-reported symptom as a diagnosis — and you will be asked to help sharpen the guideline language so the next reviewer handles that case the same way you did.
What the platform screens for
Mercor's screen is AI-led and follow-up heavy. Expect it to confirm the hard facts first — MD, two-plus years post-residency, currently seeing outpatients, US-based — and then push on whether you personally write notes rather than supervise people who do. The domain probes tend to be concrete: describe a real documentation error you caught, or explain how you would score a note that is factually accurate but clinically useless. It also tests calibration under ambiguity: whether you can distinguish a guideline gap from a genuine error, and whether you escalate rather than quietly invent a rule. Prior annotation or CDI experience helps but is not required; clear written reasoning matters more.
Logistics
- Remote, asynchronous, no set shifts — designed to sit alongside an active clinic schedule
- Minimum 10 hours/week, expected to run about three months
- Work arrives in batches through an annotation platform; turnaround windows rather than live hours
- Pay observed in the $170–190/hr band; rates are set per engagement and are not guaranteed