Senior Data Scientist - Credit Risk u0026 Provisioning Models

What You'll Do

• Develop and maintain credit risk models for Probability of Default (PD), Loss Given Default (LGD), Exposure at Default (EAD), and lifetime Expected Credit Loss (ECL) across multiple regions and products.

• Train gradient boosting models (LightGBM, XGBoost) for credit risk prediction with rigorous calibration, backtesting, and out-of-time validation.

• Design and implement vectorized models for computing forward-looking lifetime ECL estimates, incorporating macroeconomic scenarios and discounting.

• Perform feature engineering on credit datasets including payment behavior, delinquency patterns, bureau credit scores, and transactional features.

• Manage the full model lifecycle using MLflow for experiment tracking, model versioning, and registry, ensuring reproducibility and complete audit trails.

• Build and maintain model monitoring to track performance, stability, and drift across markets, producing dashboards and automated alerts.

• Develop macro-overlay models that incorporate macroeconomic variables (unemployment, GDP, interest rates) into forward-looking credit loss projections.

• Support fair value estimation and coverage rate analysis for debt sale pricing and capital management decisions.

• Run end-of-month production scoring - loading trained models, scoring exposure data at scale on cloud compute, and validating ECL outputs.

• Maintain model documentation and support audit reviews, regulatory inquiries, and model validation exercises.

• Collaborate with Data Engineers to define feature requirements, validate pipeline outputs, and ensure model inputs are accurate and timely.

• Present results to senior stakeholders including Finance leadership, auditors, and regulatory reviewers.

Who you are

• 3+ years of experience in a Data Science, Quantitative Analyst, or Credit Risk Modeling role.

• Strong Python skills for modeling, analysis, and production code (pandas, NumPy, scikit-learn).

• Experience with gradient boosting frameworks - LightGBM, XGBoost, or CatBoost.

• Solid statistical foundations - probability theory, hypothesis testing, regression, time series, survival analysis, or transition matrices.

• SQL proficiency - complex analytical queries on a data warehouse for feature extraction, validation, and ad-hoc analysis.

• Model lifecycle experience - training, hyperparameter tuning, validation, deployment, and monitoring.

• Experience with experiment tracking tools such as MLflow, Weights u0026amp; Biases, or similar.

• Strong communication skills - ability to explain model behavior, limitations, and results to non-technical stakeholders.

Awesome to have

• Credit risk modeling experience - PD, LGD, EAD, transition matrices, vintage analysis, or roll-rate models.

• IFRS 9 / CECL knowledge - staging criteria, lifetime vs. 12-month ECL, forward-looking adjustments, macroeconomic overlays.

• Familiarity with model interpretability techniques (SHAP, feature importance, partial dependence plots).

• Experience with Bayesian optimization for hyperparameter tuning.

• Exposure to Numba or vectorized computation for high-performance model calculations.

• Familiarity with fair value or pricing models for consumer credit portfolios.

• Understanding of cloud infrastructure (AWS S3, Batch, Docker) for model deployment and scoring.

• Background in fintech, banking, or consumer lending.