Data Scientist, Pipe

• 3-5+ years building ML models, including training, evaluation, and deployment
• Fluency in Python and the standard stack (NumPy, Pandas, scikit-learn, etc.)
• Working comfortably with agentic coding tools (Claude Code, Cursor, or similar) and standard dev tooling (GitHub, VS Code)
• Strong fundamentals in probability, statistics, and machine learning
• Clear written and verbal communication; comfortable collaborating across functions and surfacing progress, tradeoffs, and results
• Bachelor's in CS, Applied/Financial Math, Statistics, Economics, or a related technical field. Master's is a plus, not a requirement
• Bonus: prior experience in credit risk modeling, underwriting, or adjacent risk decision sciences (insurance pricing, actuarial, fraud) is a strong plus

Desirable

• We're looking for a hands-on Data Scientist who treats AI tooling as a core part of the craft. You'll work in an agentic development environment, such as Claude Code, Cursor, and similar, to build, ship, and iterate on statistical and ML models faster than a traditional DS workflow allows.
• The problems are real: credit underwriting, risk management, and cash-flow forecasting for SMBs across five countries. You'll learn a tremendous amount about fintech, and there's a significant career runway at Pipe
• Partner with senior scientists to forecast cash flows and other drivers of business health using statistical and ML modeling
• Design and build models that make our capital products better for customers and deliver positive return assets. Sharper pricing, lower losses, growth and offers that fit how SMBs actually run
• Proactively mine our datasets for insight, and prototype new models that move the needle for customers and the business
• Own model outcomes end-to-end. Not just deployment bespoke to the platforms Pipe serves (e.g., UberEats and Housecall Pro), but ongoing performance, residual analysis, drift detection, and the judgment calls about when a model needs to be retrained, replaced, or retired. You'll work closely with product, risk peers and engineering to make this real in production