Principal Data Scientist – Machine Learning & AI

Job description

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 (partially or fully qualified) 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. Why Accelerant? You'll have the opportunity to work on technically challenging problems that span the insurance value chain. Here you'll find: Diverse quantitative challenges across various domains The freedom to explore the rapidly evolving ML & AI landscapes from gradient boosting and deep learning to foundation models and agentic systems, while remaining grounded in rigorous experimentation and measurable business impact A collaborative team of data scientists, engineers, actuaries, underwriters, and product managers who enjoy solving difficult problems together About Accelerant Accelerant is a data-driven risk exchange connecting underwriters of specialty insurance risk with risk capital providers. Accelerant was founded in 2018 by a group of longtime insurance industry executives and technology experts who shared a vision of rebuilding the way risk is exchanged – so that it works better, for everyone. The Accelerant risk exchange does business across more than 20 different countries and 250 specialty products, and we are proud that our insurers have been awarded an AM Best A- (Excellent) rating. For more information, please visit .