Senior Credit Risk Data Scientist

The Sr. Credit Risk Data Scientist plays a central role in how Oze lends, owning the machine learning models that decide who gets credit, how much, and on what terms across our bank partners. The role is about tailoring and strengthening Oze's base credit model for new use cases and markets, and making sure every decision it makes is explainable and fair. Job Responsibilities Tailor and specialize Oze's base credit models for new use cases, customer segments, and markets as they launch through Oze Embed and Oze Originate, building use-case-specific layers rather than rebuilding the core each time Strengthen the base model by improving its predictive power, recalibrating and retraining it, and expanding its feature set to lift approval rates while holding or lowering loss rates Build explainability into every credit score so customers, partner credit committees, and regulators can understand the reason behind each decision, and support adverse-action explanations and fairness across the businesses we serve Engineer features from Oze's own business and behavioral data alongside alternative data such as mobile money, telco signals, repayment history, bank statements, and credit bureau data, including using LLMs and generative AI to turn unstructured business records into model-ready features Solve for thin-file and new-to-credit customers when entering new segments or markets, including reject inference and bootstrapping where history is limited Extend the credit lifecycle on top of the base model, covering limit assignment, risk-based pricing, behavioral scoring, early-warning and default prediction, and fraud signals Work directly with partner banks' credit and risk teams to deploy, validate, and tune models inside Embed and Originate, and explain model logic in terms they and their regulators can act on Run champion-challenger and A/B tests so model and policy changes are proven before they roll out Own model governance, including documentation, validation, monitoring for drift, and fairness checks, so models stay accurate and defensible as conditions change across markets Partner with the engineering team to take models into production, and build the portfolio analytics our partners rely on, from vintage analysis and roll rates to loss forecasting Minimum requirements and skills: A strong passion for closing the small business credit gap with and for MSMEs across Africa Proven experience building and maintaining credit or risk models in production, including logistic-regression scorecards (WOE/IV binning) and gradient boosting (XGBoost or LightGBM) Experience adapting, recalibrating, and improving existing models and transferring them across segments or markets, not only building from scratch Hands-on experience with model explainability (SHAP or similar) and the ability to translate model outputs into reasons that customers, credit committees, and regulators accept Strong Python (pandas, scikit-learn) and fluent SQL on large data sets A solid grounding in statistics and probability, and the judgment to know when a simple, explainable model beats a complex one Fluency in credit risk concepts such as PD, LGD, EAD, Gini, KS, AUC, vintage analysis, roll rates, and reject inference Awareness of responsible-lending principles and data-protection regimes across our markets, such as POPIA and the Nigeria Data Protection Act Familiarity with African data ecosystems, including mobile money, telco data, and the realities of patchy or absent credit bureaus Comfort working alongside engineers to get models into production; you understand what makes a model deployable and monitorable, while our engineering team owns the infrastructure Ability to work flexibly and efficiently on a cross-cultural team in a fast-paced environment Excellent verbal, visual, and written communicator Experience with LLMs and generative AI for extracting features from unstructured data is a plus Direct experience in African or emerging-market lending, embedded finance, or micro and nano lending is a plus Experience integrating credit bureau data or building IFRS 9 ECL or provisioning models is a plus