Principal or Senior Research Scientist, Causal and Explainable Credit Risk

ABOUT THE ROLE This is a founding research position within a new group building the next generation of consumer credit scoring from the ground up. You will own the causal inference and explainability agenda, ensuring that novel foundation model architectures can produce credit decisions that are defensible under FCRA, ECOA, and model risk management review. The work sits at the intersection of cutting-edge ML research and the regulatory realities of consumer lending. WHAT YOU'LL DO - Lead causal inference and counterfactual explanation research, including deriving actionable, stable reason codes from representation-learning models and estimating treatment effects of consumer actions on future risk. - Design the end-to-end model evaluation framework covering discrimination, calibration, economic-regime stability, fairness, and disparate-impact testing. - Build and own credit-risk modeling components, including default, recovery, LGD, and account-level hazard models, downstream of a foundation model. - Translate regulatory and model-governance requirements into concrete architectural and training constraints for the broader research program. - Author methodology documentation and present findings to lender model risk teams and regulators. WHAT WE'RE LOOKING FOR - 8 to 10 or more years of experience, with at least 5 years building and shipping credit-risk models in production at a bank, card issuer, or advanced fintech lender. - Applied causal inference depth using methods such as double ML, uplift and CATE modeling, instrumental variables, staggered difference-in-differences, or synthetic control on real business decisions. - Direct experience with explainability for regulated credit decisions, including adverse action reason codes, SHAP-based attribution, and counterfactual explanation methods, with a clear-eyed understanding of where each approach breaks down. - Working knowledge of FCRA, ECOA, Regulation B, and model risk management frameworks as they apply to credit model design. - Proficiency in Python; hands-on experience with causal inference tooling, SHAP, survival analysis, and PyTorch. - Experience designing and validating default, recovery, LGD, or account-level hazard models. - A quantitative PhD in mathematics, economics, or a related empirically focused field is strongly preferred, as is experience presenting to regulatory or academic audiences. - Familiarity with transformer or other deep learning architectures applied to credit or transaction sequence data is a plus. LOCATION Hybrid role based in San Francisco, California. Visa sponsorship is not available.