Senior Analyst — Credit Risk & Predictive Underwriting Models in London

Glow Financial Services is a consumer device financing company operating across the UK, US, Canada, and Australia.

We partner with leading telecom carriers and device manufacturers to make technology affordable, financing devices for millions of consumers — including many that traditional credit scores overlook.

We are investing heavily in the data and modeling capability that sits behind every financing decision: moving from loan-level scorecards toward customer-level, machine-learning-driven risk models that learn directly from detailed credit, payment, telecom, and device data.

This role is central to that build.

About the Role

We are hiring analysts to help design, build, validate, and monitor the next generation of underwriting and decision-support models.

You will work at the intersection of credit risk, data science, and commercial strategy — turning rich, granular data into more precise, fair, and inclusive lending decisions.

You will work with modern predictive modeling techniques, from gradient-boosted trees to representation learning, and help extend them across our four markets.

This is a hands-on, high-impact role reporting into the credit and decision analytics function within the Chief Commercial Office.

What You Will Do

• Build predictive models — develop, test, and refine models for underwriting decision support (approve/decline, pricing, deposit and affordability), using techniques ranging from regression and gradient-boosted decision trees (XGBoost/Light GBM) to neural networks and modern representation-learning approaches.
• Engineer customer-level datasets — construct point-in-time, leakage-free training data by linking loan, application, credit bureau, telecom, and device records to a stable customer key over time.
• Work with raw credit data — use record-level bureau data (tradelines, month-by-month payment histories, inquiries) rather than only summary scores and bands, to surface signal that compressed scores discard.
• Support compliant decisioning — develop and test model explainability and adverse-action reason codes so models can be used for both approvals and declines.
• Validate and monitor — assess model discrimination, calibration, and stability; support ongoing performance monitoring and re-development cycles.
• Strengthen fair lending — contribute to fair-lending and disparate-impact testing across jurisdictions.
• Document to standard — produce model documentation that meets model-risk-management and regulatory expectations (ECOA/Reg B, FCA/Consumer Duty, and equivalents in Canada and Australia).
• Partner across teams — work with credit, product, engineering, and capital markets colleagues to translate model outputs into decisions, pricing grids, and portfolio economics.
• What You Will Bring
• Bachelor's or Master's degree in a quantitative discipline (statistics, economics, mathematics, computer science, engineering, or similar).
• 5+ years in credit risk, data science, or quantitative analytics — ideally in lending, fintech, or financial services.
• Strong statistical foundation — solid command of probability and statistical inference: regression analysis, hypothesis testing, resampling, and a clear understanding of the bias–variance trade-off and the assumptions behind each method.
• Advanced analytical modeling — hands-on experience across a range of supervised techniques — logistic and regularized regression, decision trees, and ensemble methods such as gradient-boosted decision trees (XGBoost/Light GBM) and random forests — together with a working understanding of neural networks and deep learning.
• Strong Python (pandas, scikit-learn) and SQL, able to take a model from data preparation through training, validation, and evaluation (ROC-AUC, precision-recall, calibration, and model stability).
• Experience building point-in-time or leakage-free datasets from transactional or event-level data.
• Clear written and verbal communication, with the ability to explain technical work to non-technical stakeholders.
• Nice to Have
• Experience with consumer credit bureau data at the record/tradeline level.
• Deep learning at depth — hands-on experience with deep-learning frameworks (Py Torch or Tensor Flow) and with sequence models or transformer architectures.
• Experience with survival / time-to-event modeling (e. g. accelerated failure time).
• Working knowledge of credit regulation and model-risk management across the UK, US, Canada, or Australia.
• Experience using alternative data (telecom, utility, or device signals) to underwrite thin-file or no-score populations.
• Exposure to cloud ML infrastructure and GPU-based model training.
• #J-18808-Ljbffr