Principal Data Scientist - Machine Learning & AI

About the role

We're looking for a Data Scientist to develop machine learning and AI systems that improve decisions across pricing, underwriting, portfolio management, operations, and claims. You'll work across structured data, text, documents, and external data sources, applying statistical modeling, modern machine learning, AI and agentic workflows to solve challenging real-world problems.

The foundation of this role is serious quantitative modelling. We care about calibration, not just discrimination. We validate out of time and worry about leakage and drift. We quantify uncertainty and can tell you when a model should be trusted, when it shouldn't, and why. LLMs and agentic systems are a force multiplier on all of that and we measure those systems the way we'd measure any other model: on data they haven't seen, against a sensible baseline, with honest uncertainty around the result. You don't need an AI background to join us; you do need genuine enthusiasm for working this way.

This is not a reporting or dashboard role. You'll work on ambiguous, high-impact problems where you'll be expected to identify the right approach, build production-ready solutions, and measure the business impact of your work.

If you enjoy messy data, difficult prediction problems, and building intelligent systems that make real-world decisions better, you will be a good fit.

What You'll Work On

Our team tackles a broad range of machine learning and AI problems. Depending on business priorities, you may work on projects such as:

• Predictive modeling for pricing, underwriting, claims, catastrophe risk, and portfolio management
• Classification, ranking, matching, recommendation, and anomaly detection systems that improve business decision-making
• Information extraction from documents, emails, forms, and other unstructured data using modern AI techniques
• Entity resolution, data enrichment, and building high-quality datasets from noisy or incomplete information
• Design AI systems that automate analytical and decision-making workflows end to end. Build the measurement that tells us whether they genuinely outperform what they replace
• Building production feature pipelines, model inference services, and evaluation frameworks
• Collaborating with engineers, actuaries, underwriters, product managers, and business leaders to turn ambiguous questions into scalable machine learning solutions

What We're Looking For

You likely have experience with many of the following

• A strong quantitative foundation: statistics, probability, optimisation, or applied mathematics
• Sound modelling judgement - you know what it takes for a model to hold up in the real world, not just on a validation set
• Strong programming skills
• Real willingness to work with LLMs and agentic AI as everyday tools, wherever your background sits today
• Clear communication with both technical and non-technical audiences - you can explain a lift curve to an underwriter and a shrinkage prior to a statistician

Bonus Points

Experience in one or more of the following is especially valuable

• Track record with LLM-powered applications or AI agents, especially if you've done the unglamorous work of proving they perform
• Depth in the statistical toolkit beyond supervised prediction: hierarchical models and shrinkage estimation, causal inference and experimentation, survival analysis, extreme value theory, or demand and elasticity modelling
• Insurance domain knowledge: pricing, reserving, claims, underwriting, or distribution

Actuarial background or qualifications

• Experience in regulated industries where model governance and explainability matter
• ML engineering experience: taking models from research code to production services, or building the tooling and frameworks that help others deploy
• Cloud and infrastructure skills: AWS, Azure, or GCP; containers and orchestration; APIs and data pipelines built with cost, latency, and reliability in mind
• MLOps in practice: experiment tracking, model monitoring, automated retraining, and CI/CD for models and agent

Team Context

You'll join a lean, senior team with low bureaucracy and high autonomy. We're investing heavily in agentic AI as the next evolution of how a quantitative team operates, and you'll help shape that direction from the start.