Sr Data Scientist

Role: Senior Data ScientistComp Range: 150K-200K baseNo 3rd party/C2C Client is re-writing the risk model for the non-QM mortgage market -combining data analytics, proprietary risk intelligence, and Loan Defect Insurance to turn mortgage manufacturing risk into quantifiable, insurable outcomes for lenders, investors, and RMBS issuers. Client AI (CLIENT) is the analytical engine powering this platform, built on neural network triage, hazard-based pricing, and AI-driven defect detection. The Senior Data Scientist owns the science behind this platform;building and continuously improving the full model suite across triage, pricing, defect detection, compliance verification, fraud analytics, and claims modeling. In this role you will: · Build, train, and iterate CLIENT's core ML models: pBad triage, hazard-based risk pricing, and defect detection modules covering appraisal, income recalculation, and ATR compliance · Build and maintain the servicer compliance verification module - programmatically checking servicer adherence to guidelines and flagging material deviations · Design and implement cross-loan fraud pattern detection using anomaly detection and network-based methods to identify coordinated misrepresentation across loan pools · Build claims and damages analytics -estimating expected loss severity, cure rates, and recovery timing calibrated against realized outcomes · Maintain mark-to-market LTV monitoring using live HPI feeds, producing collateral exposure updates and early-warning signals · Maintain formal model governance documentation: assumption logs, backtesting results, out-of-sample validation, and performance monitoring cadence The Ideal Candidate: · Equally fluent in model science and mortgage credit -neither alone is suƯicient · Holds model outputs to the standard an institutional counterparty would apply · Brings intellectual rigor to messy, real-world loan data Basic Qualifications: · 6+ years of quantitative modeling with at least 3 years in mortgage credit risk, structured finance, or specialty insurance · Supervised ML models built, deployed, and reviewed by external auditors · Strong Python with scikit-learn, XGBoost, or PyTorch Preferred Qualifications: · Non-QM mortgage underwriting, defect taxonomy, and R&W risk frameworks · Survival models, hazard models, or actuarial loss estimation methods · Fraud detection using anomaly detection or network-based methods on financial data · Servicer compliance requirements and non-QM due diligence data sources · Model documentation packages reviewed by institutional counterparties or regulators · Claims analytics, loss given default modeling, or insurance loss reserving Tech: Python · scikit-learn / XGBoost / PyTorch · SQL · lifelines or equivalent · MLflow · Git