Principal Data Scientist

About the role

We are seeking a Principal Data Scientist to join Pedestal Health's Quantitative Sciences (QS) organization. In this role, you will build the infrastructure and methods that make AI-assisted analytics and real-world data quality work fast, scalable, and genuinely trustworthy. Your work will span two connected areas. The first is AI-enabled analytics: building and maintaining AI tooling and workflows that let our team identify cohorts, review analysis code, and carry out recurring analytic work faster and more consistently. The second is AI-enabled data quality: designing automated and agentic approaches that scale across schemas, sites, and data refreshes, and that surface issues before they reach an analysis or a client. This is a role about building capability, not about producing analyses. You will design, build, and scale the internal tools that change how our teams work with real-world data — and you will own them as products, with users, versions, quality standards, and a roadmap. This is a hands-on role that combines individual technical contribution with technical leadership, including mentorship and review of other data scientists' work. You will partner closely with Product, Engineering, Medical, and Commercial teams, and report to the Head of Quantitative Sciences.

Responsibilities

• Build, maintain, and improve AI-powered tooling that lets the team generate commercial cohort counts and conduct feasibility reliably and repeatably.
• Extend the same approach to other recurring analytic work, including generating and reviewing analysis code, and supporting protocol and analysis plan development.
• Gather requirements from the internal teams who depend on these tools, treat them as users, and iterate on real feedback rather than assumed needs.
• Own what keeps this tooling trustworthy over time (how it is tested against known-correct results, how updates are validated before release, and how performance is monitored as the underlying data evolves), and where human review remains mandatory.
• Scale adoption across the team through documentation, training, onboarding, and hands-on enablement.
• Help define the guardrails for AI use in client-facing work: what data may be used, how outputs are reviewed and by whom, and how provenance is recorded.
• Design the infrastructure that tells us whether our data is trustworthy, before anyone else has to find out.
• Rethink how our data quality checks are built and run, so that assessing a new source or a refreshed schema no longer means redoing the work each time.
• Move quality assessment beyond manual review and spreadsheet outputs, toward an automated approach with a durable