Epidemiologist — Patient Population Sizing

$150–175/hrrate as listed

Mercor is hiring a US-based epidemiologist to build and critique rubrics that judge how AI systems size patient populations for commercial drugs and development-stage programs. The pilot runs roughly 10–20 hours over one to two weeks, with observed rates of $170–270/hr and the possibility of continuing as the work scales.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work involves

The core task is epidemiological judgment applied to AI output. You will be handed patient-population estimates — sometimes model-generated, sometimes drawn from analyst-style write-ups — for a specific drug or indication, and asked to determine whether the estimate holds up. That means walking the funnel: prevalence or incidence source and vintage, diagnosis rates, treatment rates, line-of-therapy restrictions, biomarker or subpopulation gating, and what actually remains as an addressable population. You will flag double-counting across indications, silent extrapolation from one geography to another, prevalence estimates applied to incident-only settings, and claims-based denominators that quietly exclude uninsured or Medicaid populations.

The second half of the work is rubric authoring and critique. You write the criteria another evaluator — or a model — should use to score a population estimate: what counts as adequate sourcing, when a range is more honest than a point estimate, how much weight a missing sensitivity analysis should cost. You will also review rubrics written by others and argue when they reward surface-level citation over defensible method.

What the platform screens for

Real fluency with claims, EHR, and registry data — specific datasets you have used, and the known biases of each
Ability to reconstruct a diagnosed → treated → addressable funnel out loud and name the assumption at each step
Judgment under ambiguity: whether you can say an estimate is wrong and explain by roughly how much and in which direction
Familiarity with how pharma and biotech teams actually use these numbers in program decisions
Written clarity, since most deliverables are prose critiques and rubric text rather than code or tables

Logistics

Fully remote, US-based only, and asynchronous — you pick up batches on your own schedule with per-batch deadlines rather than fixed hours. Expect 10–20 hours total across a one- to two-week pilot. Rates in the $170–270/hr range have been observed for this listing; actual offers depend on screening outcome and scope. Strong contributors are typically invited into recurring work as the evaluation set expands.

What they look for

Graduate degree in epidemiology, biostatistics, or public health
5+ years practising epidemiology
Experience converting claims, EHR, or registry data into patient funnels
Familiarity with biotech/pharma drug-development programs and commercial indications
Ability to critique and author evaluation rubrics in clear written English
US-based and authorized to work as an independent contractor

Before you apply

Would you pass this screen?

Drawn from what this listing actually requires. Answered in your browser — nothing is sent to us, and we do not store your answers.

Are you currently located in and authorized to work in the United States?
US-based · required
Do you hold a master's degree or higher in epidemiology, biostatistics, public health, or a closely related field?
Graduate degree in epidemiology, biostatistics, or public health · required
Have you worked as a practicing epidemiologist for at least 5 years after completing your graduate degree?
5+ years epidemiology practice · required
Have you directly worked with claims data, electronic health records (EHR), patient registries, or similar real-world health data to estimate or analyze patient populations?
Real-world data experience · required
Have you worked on, consulted for, or directly analyzed drug-development programs, clinical trials, or pharmaceutical/biotech research?
Biotech/pharma familiarity · preferred

0 of 5 answered

Included with this listing

Interview prep — 10 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: Whether the candidate can articulate the real-world funnel (population → diagnosed → treated → addressable market) and knows which data sources (claims, EHR, registries) speak to each stage.
A strong answer: The answer names a concrete indication, identifies at least two distinct data sources (e.g., claims data for treatment rates, registry for diagnosis prevalence), and explains the logical chain from one population figure to the next—showing they've actually built a funnel, not just cited raw numbers.
Where people lose it: Confusing incidence with prevalence, or treating 'diagnosed' and 'treated' as interchangeable; claiming to use claims data but unable to explain what that dataset actually contains or its limitations; naming a drug without specifying the indication or the population constraint.