Data Scientist

Replit is the agentic software creation platform that enables anyone to build applications using natural language. With millions of users worldwide, Replit is democratizing software development by removing traditional barriers to application creation.

About the Role

Replit spreads through companies differently than it does through individuals. One person builds an app, colleagues start using it, a team forms, an admin rolls Replit out more widely, and eventually Sales gets involved. You'll be the data scientist for Replit's enterprise business, covering both the product and the go-to-market motion: how Replit spreads inside organizations, what keeps teams engaged after rollout, which features and signals drive expansion, and how all of that becomes pipeline, revenue, and a reliable forecast.

Experimentation is hard here. There are far fewer accounts than users, people within an organization influence each other, and admins control what their teams see. You'll figure out how to learn quickly and rigorously anyway. You'll work with Enterprise Product, Engineering, Sales, Customer Success, RevOps, and Finance, and your work succeeds when it changes a product, GTM, or investment decision.

You will:

• Define how we measure the enterprise business end to end: account-level activation, breadth (share of seats active), depth (apps built, deployed, and shared), collaboration, consumption, and net revenue retention

• Understand virality inside organizations: how Replit spreads from a first builder to teams through app sharing, invites, and viewers who become builders; quantify within-account network effects and find the levers that accelerate them

• Explain post-engagement churn: why seats and accounts that engaged at rollout go dormant, separate seat-level decay from account-level risk, and turn it into account health models that give the product and Customer Success early warning and a clear next action

• Identify the product behaviors and features that precede seat expansion and plan upgrades, and use them to build product-qualified account (PQA) models that route bottoms-up usage to Sales at the right moment

• Measure adoption and value for enterprise features — admin and governance controls, security, data connectors, collaboration, and deploying to data platforms — and use it to inform what we build next

• Design experiments for enterprise: account- and cluster-level randomization, interference within organizations, small samples and heavy-tailed account sizes, stratified and matched-pair designs, staged rollouts and long-running holdouts, synthetic control, and Bayesian or hierarchical models where data is sparse