Dermatologist, U.S. Licensed

What you would do

• Systematically examine clinical photographs to identify skin pathology and describe findings using appropriate medical terminology
• Compare AI-generated diagnostic conclusions against what an experienced dermatologist would observe and recommend
• Author evaluation rubrics that specify what constitutes acceptable versus inadequate clinical assessment in particular contexts
• Advise engineering teams on how their algorithms diverge from authentic diagnostic workflows and reasoning

Who they want

• Hold an M.D. degree with an active, unrestricted United States medical license
• Possess minimum five years of post-graduate dermatology training and clinical practice (current or prior experience)
• Demonstrate capacity to articulate complex skin conditions with medical precision and accuracy
• Proficiency in English language communication for written and verbal collaboration

Main skills

What the interview asks about

1. 1.Dermatological image interpretation

   This role fundamentally depends on rapid, accurate recognition of skin conditions from photography; the interviewer needs to confirm you can reliably distinguish common mimics and identify subtle morphological features that guide diagnosis.

   For example: “You receive a portfolio of 40 images showing various papules and plaques. Walk through your approach to categorizing them into likely diagnostic groups, and describe how you would recognize which cases fall outside dermatology's domain.”

2. 2.Evaluating algorithmic reasoning

   You will regularly encounter model recommendations that diverge from clinical standards; assessing whether these errors reflect genuine gaps in training data versus bugs in feature weighting is crucial to your feedback value.

   For example: “An AI flags a lesion as high-risk melanoma, but clinical features suggest benign nevus. How would you document this error to help engineers understand where the model's reasoning diverges from clinical standards?”

3. 3.Guideline authorship and standardization

   Ambiguity in evaluation standards destroys consistency across reviewers; your ability to write explicit, unambiguous criteria ensures that other evaluators and deployed models interpret images using the same logic you do.

   For example: “Your team needs a protocol for assessing lesion morphology variation across body sites. How would you structure criteria for 'well-demarcated borders' so that different reviewers looking at elbows versus faces reach the same conclusion?”

4. 4.Technical communication under constraints

   Engineers lack medical training; your ability to translate diagnostic concepts into their terminology without losing precision determines whether they can implement your feedback effectively.

   For example: “You notice the model consistently mislabels lichen planus as psoriasis. How would you explain the distinguishing clinical features to a computer vision team, knowing they have no medical background but understand image segmentation and feature extraction?”

5. 5.Sustained accuracy in high-volume review

   This role requires hours of visual analysis with consistent precision; fatigue and habit errors could silently introduce bias into model training data, so you must demonstrate systems for maintaining rigor over extended sessions.

   For example: “You're scheduled to review 200 images over two days for consistency scoring. Describe the checkpoints and verification steps you would build into your workflow to catch your own drifting standards mid-way through.”

A task you may get

Review a dataset of 20 dermatological images, classify each with diagnosis and confidence rationale, then compare your assessments against provided AI predictions and identify patterns where the algorithm diverges from clinical reasoning.

How to prepare

• Review recent dermatology literature on common diagnostic pitfalls and mimicry patterns to refresh your pattern recognition
• Practice documenting skin findings using structured terminology to build speed and consistency in your written assessments
• Study how machine learning engineers interpret 'accuracy' and 'precision' so you can frame clinical observations in terms they can operationalize

The facts

Pay

$270/hr

Hours

Hourly, 40 hours a week

Where

Remote

Open to

USA

Posted

9/9/2026

We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.