Health Economics Expert Model Evaluator

You will create and assess expert evaluation data for AI models. The work uses health economics and outcomes research judgment to test whether model-generated analyses are analytically sound and defensible.

• Develop realistic economic evaluations from the research question through the final recommendation.
• Write reference answers that explain the methodological reasoning behind each decision.
• Grade model outputs for analytical soundness, defensibility of assumptions, and quality of reasoning.
• Identify analyses that may mislead payers or regulators even when they appear plausible.
• Apply health economics judgment to cost, evidence, and health-outcomes reasoning.

This is a remote, asynchronous contractor opportunity with ongoing project work. The role is open across the countries listed for the project and requires clear written English.

• Time commitment: 20 or more hours per week.
• Language: Professional fluency in English.
• Education: A graduate degree such as an MS, MPH, or PhD in health economics, economics, public health, epidemiology, or a related quantitative field.
• Experience: At least three years of professional or research experience in health economics or outcomes research.
• Required background: Hands-on experience building economic models and documenting methodological reasoning clearly.
• Important expertise: Cost-effectiveness modeling, budget impact, real-world evidence, or health technology assessment submissions.
• Helpful tools: R, TreeAge, Excel, VBA, Stata, or SAS.
• Helpful data experience: Claims or electronic health record databases such as MarketScan, Optum, IQVIA, Medicare, SEER-Medicare, or Flatiron.
• Helpful submission experience: NICE, ICER, CADTH, PBAC, IQWiG, or AMCP submissions.

Apply on OpenTrain with your resume, then complete the application on the hiring site.

AI training work uses human-written examples and expert reviews to measure and improve how artificial intelligence systems respond. People with specialized knowledge are needed to judge whether model outputs are accurate, well reasoned, and suitable for real-world use.