Founding Data & Machine Learning Lead

About the role

This is Keep Company’s first dedicated data hire. It is a hands‑on build role: initially weighted toward data engineering and analytics infrastructure, with meaningful ownership of model development, evaluation, and strategy as the foundation matures. You will own how data moves through Keep Company: from source systems and product events to clean, trusted, member‑centric datasets, reporting, and early predictive use cases. You will partner closely with Product and Engineering to define the signals we should capture, build the infrastructure that makes those signals usable, and evaluate where models can improve matching, recommendations, and intervention decisions. The first phase will focus on foundations—pipelines, data modeling, instrumentation, data quality, and reusable reporting. Over time you will lead the strategy and development of models that help identify risks, improve relationship outcomes, and make more useful recommendations.

Responsibilities

• Own Keep Company’s analytical data foundation, including the warehouse, transformations, pipelines, data quality, and documentation.
• Establish MotherDuck as a reliable analytical source of truth, including dbt models, source definitions, and durable datasets for internal and in‑product use.
• Partner with Engineering to improve how product, program, survey, HRIS, import, and coach data is collected, structured, and maintained.
• Build and maintain pipelines across current inputs, including WorkOS/HRIS integrations, CSV imports, surveys, forms, and platform activity.
• Bring historical data into a clean, member‑centric structure that connects attendance, registrations, survey responses, coaching inputs, relationship activity, and outcomes.
• Define and maintain a shared taxonomy for constructs, questions, scales, programs, cohorts, and other historical data that is currently inconsistent.
• Partner with Product and Engineering to define product instrumentation: what actions, milestones, decisions, and outcomes should be captured, where they should live, and how they can be used.
• Build reusable measurement layers and reporting products rather than one‑off analyses.