Director of Modeling

About You

You think in experiments — your first instinct is "how would we test that?" You're rigorous but pragmatic, designing studies that are defensible without being precious about textbook conditions. You build teams, not just models, and you care about hiring well, setting standards, and making the people around you better. You ship production code, not just notebooks. You communicate with conviction — no jargon, no hedging everything into uselessness. And you're energized by high-stakes, messy problems; political environments are noisy, fast, and consequential, and that sounds fun to you.

About The Role

Tavern runs high-stakes experiments in complex political environments — persuasion, mobilization, media effects — where getting the model wrong has real consequences. As we scale the number of clients and campaigns we support, we need a modeling leader who can build the team, the infrastructure, and the experimental rigor to run dramatically more experiments without sacrificing quality. This role exists to turn modeling from a bottleneck into a competitive advantage. This role blends hands-on modeling with strategic leadership and cross functional coordination.

Responsibilities

• Set the vision and roadmap for data science at Tavern — what we build, what we buy, what we skip, and why.
• Build and lead a high-performing data science team, starting small and scaling with the business; hire, mentor, and set the bar for quality and rigor.
• Own a centralized, prioritized data science function that serves delivery, organizing, and product — not ad-hoc analytics scattered across teams.
• Define measurement frameworks for online organizing programs; ensure we can credibly demonstrate impact to clients and internal stakeholders.
• Establish company-wide standards for schemas, metrics, experimental design, code quality, and reproducibility — creating a shared source of truth.
• Partner with engineering and product leadership to productionize models and data products; champion a "data science as engineering" culture over bespoke one-offs.
• Translate complex methods into plain language for executives, delivery leads, and clients — surfacing tradeoffs and uncertainty, not just point estimates.
• Bring genuine technical depth — strong enough to review code, pressure-test models, and architect data infrastructure yourself; you're as comfortable as a data engineer as you are as a data scientist. But your job is to build systems and people, not just do the work.
  
  

Qualifications

We know that it’s rare to check all the boxes here, so if you check most of these you should consider applying:

• Deep expertise in causal inference and experimental design — you've designed, run, and analyzed RCTs or quasi-experiments in applied settings, not just coursework.
• Strong programming skills in Python, including scientific and statistical libraries; you write code that other people can read, run, and build on.
• Experience building production-grade model pipelines — not just training models, but deploying, monitoring, and maintaining them.
• Familiarity with survey methodology, measurement, or behavioral modeling — political experience is a plus but not required.
• Working knowledge of cloud infrastructure (GCP preferred) and comfort with modern development practices (version control, CI/CD, containerization).
• Experience managing or mentoring data scientists or applied researchers, or a clear track record of elevating the work of people around you.
• Strong data engineering skills.
• Bonus: exposure to NLP, media analysis, or treatment effect heterogeneity in applied contexts.
  
  

Benefits

• Premium health insurance
• Unlimited PTO
• Office closed for all federal holidays
• 401k match
• Equity
  
  

Tavern Research is an equal opportunity employer. We celebrate diversity and are committed to creating an inclusive environment for all employees.