$60–180/hr · Mercor · Part time, 40 hours a week
A clinician or health professional contributes medical expertise to AI training, writing cases and grading model responses on clinical reasoning.
What you would do
- Write realistic clinical cases from your own practice with complete presentation, workup, and clinical decision points
- Grade AI model output on diagnostic reasoning, assessing whether differentials are ordered logically
- Evaluate whether clinical guidelines appropriately apply to specific patients or whether modifiers are needed
- Identify where textbook answers differ from defensible real-world practice given patient constraints
- Flag ambiguous instructions and contribute feedback for AI training material improvement
Who they want
- Valid licensure and hands-on experience practicing in a healthcare field
- Clear written communication skills for explaining clinical reasoning
- Comfort with ambiguous tasks and willingness to raise unclear instructions
- Experience spanning internal medicine, surgery, psychiatry, nursing, pharmacy, or public health
- Ability to articulate both standard protocols and pragmatic real-world deviations
What the interview asks about
1.Case design with clinical realism
AI learns from examples; your cases must reflect actual decision-making complexity and constraints you encounter.
For example: “You write a case of a 68-year-old with hypertension, diabetes, CKD, and chest pain in a rural clinic with limited imaging. How do you structure the presentation and reference answer to capture both evidence-based reasoning and practical constraints?”
2.Diagnostic reasoning assessment
Evaluating differential diagnosis requires you to judge not just correctness but the order and logic of thinking.
For example: “An AI model generates a differential for a patient with dyspnea that includes pulmonary embolism as the third choice, when your clinical sense says it should rank higher given the risk factors. How do you document this assessment?”
3.Guideline interpretation boundaries
Guidelines are tools, not rules; you identify when they fit and when patient factors mandate deviation.
For example: “A guideline recommends invasive monitoring for septic shock management, but your patient refuses, and the ICU is full. How do you grade a model's adherence to guidelines when the 'correct' answer requires nuanced judgment?”
4.Comorbidity interaction analysis
Textbook scenarios assume single-organ disease; your expertise is knowing how comorbidities change the plan.
For example: “You're grading a response for a hypertensive diabetic with recent stroke considering a statin. The model says to use standard dosing, but your patient has severe myopathy. How do you explain where the model missed clinical context?”
5.Balancing evidence and pragmatism
Frontier AI learns from examples of when and why practitioners deviate from guidelines; this is subtle but important.
For example: “Evidence-based practice says to screen all diabetics for proteinuria annually. Your clinic has no lab capacity this week and the patient is stable. You defer screening two weeks. How do you document this deviation and explain it as defensible?”
A task you may get
Write a clinical case with diagnostic question from your specialty. Include your reference answer explaining clinical reasoning with one real-world constraint that modifies standard protocol.
How to prepare
- Select 2-3 memorable cases from your practice that illustrate good diagnostic thinking or real-world tradeoffs
- Write out your clinical reasoning for one recent decision in detail, noting where evidence supports you and where you relied on experience
- Identify one clinical guideline in your field and a scenario where you would deviate from it - prepare to explain why
- Practice writing clinical explanations aimed at both specialist and non-specialist audiences
The facts
- Pay
- $60–180/hr
- Hours
- Part time, 40 hours a week
- Where
- Remote
- Field
- Medicine
- Role type
- Talent network
- Posted
- 9/15/2026
We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.