Data Scientist Manager

About Us: Perpay is a certified B Corp and Philadelphia’s most impactful growth-stage startup. We are driven by a mission to significantly improve the financial stability of everyday Americans. For the past decade, we have established strong product-market fit and a profitable, efficient operating model across a suite of products, positioning Perpay as the premier financial partner for consumers with subprime credit. With over 500,000 customers who have utilized more than $1 billion in spending power, we are at a pivotal moment. We are scaling our operations, building new offerings, and deepening our impact. We are looking for teammates eager to join us on this journey. Our venture partners include First Round Capital and L Catterton. Products we’ve built to make an impact: Perpay Marketplace: Combines interest-free payments and modern e-commerce to reduce cost of ownership and promote healthy repayment behavior. Perpay+: Leverages Marketplace repayment history to help members monitor and build credit with all 3 credit bureaus. Perpay Credit Card: Expands access to the flexibility and benefits of a World Mastercard by removing common barriers like high security deposits and low approval odds. Our team thrives on in-person collaboration, operating from our unique center-city Philadelphia office. This comfortable "home away from home" space offers river views and fosters [... source excerpt omitted ...] drives Perpay's most consequential decisions: credit decisioning, loss forecasting, marketing-mix attribution, product experimentation, and the ML systems that sit in front of our customers in real time. This year, with the credit portfolio scaling and our modeling needs getting heavier, focus areas include owning the data science side of the risk decisioning service redesign, expanding our card-portfolio modeling, deepening our use of LLMs in both internal workflows and customer-facing surfaces, and tightening the feedback loops between our credit-reporting strategy and the data that informs it. Data Science partners directly with Engineering, Risk, Marketing, Merchandising, and [... source excerpt omitted ...] ure and shared problems. Our data science culture leans toward end-to-end ownership: the person who designs a model should be the one who scopes it with stakeholders, ships it to production, and stays close to how it performs once it is live. We invest in rigor where rigor matters and resist the urge to over-engineer where it does not. We are comfortable being challenged on our work and comfortable challenging back, because the alternative is shipping models that look right and are not. The stack: Python everywhere, with the standard data science toolset (scikit-learn, pandas, NumPy, matplotlib, statsmodels) and Bayesian tooling (PyMC) on the projects that need it. Models are depl