Staff Data Scientist

Who you are

• Significant hands-on experience in data science or quantitative decisioning, with meaningful time in consumer credit, lending, fintech, or another risk-heavy environment where model quality has direct financial consequence
• Strong fluency in Python and SQL for model development, feature engineering, analysis, and production-grade investigative work
• Deep track record building and evaluating predictive models in production: feature pipelines, validation design, calibration, monitoring, and full lifecycle management
• Solid grounding in credit risk concepts and portfolio metrics - probability of default, loss behaviour, bureau and alternative data - and the practical judgement to apply them in a fast-moving startup
• Experience partnering with engineering teams to deploy decisioning or ML systems into live products, not just offline research environments
• A clear record of turning analytical work into measurable commercial or risk outcomes: improving approval quality, reducing losses, increasing conversion responsibly, or sharpening collections strategy
• Confident operating at staff level in ambiguity: setting direction without fully defined scope, influencing senior stakeholders, and improving systems through judgement
• Experience with bureau data integrations or alternative data sources in emerging or frontier markets
• Familiarity with explainability frameworks (SHAP, LIME) and regulatory expectations around model governance in consumer lending
• Prior experience in a high-growth fintech or startup where you have had to balance scientific rigour with speed of iteration

What the job involves

• LemFi's Credit function is one of the most consequential bets we are making as a business
• We lend to customers the traditional financial system has largely ignored, and getting that right - at scale, responsibly, across multiple markets - demands serious data science
• As Staff Data Scientist for Credit, you sit at the centre of that effort: partnering across Credit Risk, Product, Engineering, Analytics, and Commercial to shape how we underwrite, monitor, and grow lending
• Your models will run in production
• Your recommendations will land in the room where senior leadership decides risk appetite and market expansion
• This is not a research role
• It is a high-ownership, high-consequence position for someone who wants their work to show up in portfolio performance, approval quality, and customer outcomes - not just in a notebook
• You will set the technical and scientific standard for credit decisioning at LemFi, mentor the data scientists and analysts around you, and help the business grow its credit products responsibly
• Own end-to-end data science strategy for Credit, from underwriting and pricing to collections optimisation and lifecycle decisioning, surfacing the highest-value opportunities and driving them to production
• Design, train, validate, and iterate on predictive models (probability of default, loss estimation, segmentation, fraud interaction) using machine learning and statistical methods calibrated to a regulated, real-world environment
• Work with Data Engineering and Software Engineers to productionise models cleanly: features, decision logic, versioning, and monitoring all reliable and scalable before anything goes live
• Build robust monitoring frameworks covering model performance, drift, fairness, and operational impact - with a high bar for explainability and governance that satisfies both internal and regulatory scrutiny
• Partner with Product and Credit leaders to translate ambiguous business problems into measurable hypotheses, experiment designs, and decision proposals - using AI tooling to move fast without sacrificing rigour
• Lead deep-dive portfolio analyses into repayment patterns, loss drivers, and customer segments, converting findings into concrete recommendations for senior stakeholders
• Act as technical mentor and thought partner to analysts and data scientists, raising the bar on problem framing, methodology, and the quality of decisions the team produces