Senior Data Scientist — Actuarial Modelling

Oliver Brandt

Contract-to-hire · Senior · London

This role is no longer accepting applications

Similar open roles you can apply to now:

About the role

We are a specialist motor and home insurer operating in the UK personal lines market. We write approximately 1.4 million policies per year. Our pricing function has historically been led by qualified actuaries using traditional GLM-based approaches. Over the last three years we have been building a parallel capability in ML-based pricing — gradient boosting models for frequency and severity separately, with ensemble approaches for the final technical price. The two approaches coexist in a productive tension. Our actuaries are excellent. They are also, by training and temperament, sceptical of models they cannot interpret. The Senior Data Scientist we are hiring will need to produce models that are accurate enough to improve our loss ratios and explainable enough that a Fellow of the Institute of Actuaries is willing to vouch for them to the pricing committee. That is a specific challenge. It rewards people who are good at both the science and the communication. We are starting with a six-month contract because we want to make sure the technical approach and the internal fit are right before committing to a permanent hire — which is our strong intention if both are.

Responsibilities

Build and validate ML pricing models for motor and home insurance, with documentation suitable for actuarial review and regulatory audit
Conduct comparative analysis between ML and traditional GLM approaches — loss ratio impact, Gini coefficient improvement, and stability under portfolio stress
Collaborate with qualified actuaries to integrate ML outputs into the pricing committee sign-off process
Develop model monitoring and drift detection for production pricing models
Contribute to the team\'s technical documentation standards and model governance framework

Requirements

6+ years of data science with at least 3 years working on insurance pricing, reserving, or risk modelling in a regulated environment
Python at a senior level — production-quality modelling code, not just notebooks, with reproducibility and documentation standards appropriate to actuarial sign-off
Gradient boosting frameworks — XGBoost or LightGBM — for frequency and severity modelling on large insurance datasets
Statistics at depth: GLM theory, tweedie distributions, exposure-weighted modelling, and the confidence to defend methodological choices to actuarial challenge
SQL for extracting and transforming policy, claims, and exposure data from our data warehouse
Scikit-learn for evaluation pipelines, feature selection, and baseline model comparison
Familiarity with FCA pricing regulations and the actuarial validation requirements in UK personal lines is a significant advantage

Benefits

$120,000 – $150,000 annualised contract rate — strong intention to convert to permanent
6-month initial contract with clear conversion criteria agreed at start
Hybrid — two days in our Bristol office, three days remote
Actuarial peer group that will challenge your methodology rigorously — which you will find either stimulating or uncomfortable, and we think you already know which
Exposure to real P&L impact from model changes