Senior Data Scientist, Decision Science

Overview

In this role you apply classical ML and decision science to improve lead quality and traffic performance, linking model outcomes to commercial KPIs. You will work with AI engineers, product managers, and operators to deploy predictive models and policy learning that respect existing rules while driving evaluation-based improvements. The role combines model development, validation, and off-policy learning to responsibly scale AI-driven actions. You help translate business rules into actionable decisions and measurable impact.

Pay / Benefits

• Summer Fridays
• 25 days holiday + 8 bank holidays (+1 day each year up to 30)
• Hybrid working – 3 days in office
• Life Assurance and Income Protection
• Health, dental, and vision coverage
• Pension and Salary Sacrifice Scheme

Responsibilities

• Develop and deployed ML models for lead quality and traffic performance (classification, ranking, anomaly detection, forecasting).
• Evaluate, monitor, and manage model performance with offline validation and live drift checks, collaborating with the platform team for deployment on SageMaker.
• Demonstrate commercial impact through A/B testing and KPI measurement.
• Frame decisions by turning business rules into context, actions, and rewards; establish baselines.
• Research and apply policy learning methods (contextual bandits, safe exploration, off-policy evaluation) for live decisions.
• Quantify agreement with experts, override rates, and establish confidence guards before autonomous actions.
• Support hand-over with evidence and guardrails; facilitate sign-off by business stakeholders.
• Contribute to architecture and AI-native delivery, ensuring explainability and maintainability of models.

Key requirements

• 5+ years as a data scientist or in a similar applied ML role with production models.
• Statistics and experimentation including hypothesis testing, A/B testing, power and sample sizing.
• Classical ML: classification and ranking with scikit-learn and XGBoost or LightGBM; rigorous evaluation (validation design, calibration, leakage).
• Decision science: applied work with contextual bandits, causal inference or uplift modelling; ability to work independently.
• Strong Python and SQL; Git and code review.

• Candour and clear communication
• Delivery ownership and accountability
• Ability to teach and uplift teammates
• Python
• SQL
• Git

…