Data scientist in insurance underwriting

This role puts you at the heart of the business, working side by side with our underwriters to develop solutions that really matter. Bringing together data science and AI engineering, it has a clear goal: making a measurable difference to underwriting. As one of the first hires into our new Underwriting Analytics function, you will be working in one of the most exciting parts of the business, with a real opportunity to make an impact.

• Spend time with underwriters, portfolio managers and product heads to understand their challenges first-hand, and agree which problems are worth solving and what success looks like.
• Use analytics and technology to solve each problem, choosing the simplest approach that works, whether analysis, automation, an AI workflow, a statistical or machine learning model, or a third-party product.
• Use large language models, document AI and automation to turn unstructured information, such as submissions, emails and documents, into structured data, and to take repetitive manual effort out of underwriting processes. The data these tools capture will power the analytics that follow.
• Develop working prototypes in collaboration with underwriters, and use their feedback to shape solutions that work in practice.
• Pass successful prototypes to our central data and technology teams with clear requirements, results and success measures, then support them through delivery and adoption.
• Build statistical and machine learning models, such as segmentation, triage and propensity models, that help underwriters select risks, manage referrals, refine appetite and monitor portfolio performance.
• Analyse broker flow, quote-to-bind conversion and portfolio performance, and track changes in rates, terms and conditions to spot shifts in the market early. Share what you find in plain language, so underwriting teams and senior management can see where to focus and when to act.
• Shape how the new function grows, and keep up with developments in data science and AI so we can bring the best ideas into underwriting.

• You have applied statistical and machine learning methods, to real business problems, and can design analyses and experiments that measure impact.
• You write clean, well-structured Python and SQL, use version control and testing as standard, and are comfortable working with large, imperfect datasets.
• You have built applications or prototypes using large language models, including prompt design, structured outputs and retrieval, and evaluated how well they work.
• You have turned unstructured information, such as documents, spreadsheets or emails, into structured data, and checked how accurate the results were.
• You have taken a business problem from first conversation to a working prototype that people tested or used, and measured the difference it made.
• You have handed over models or prototypes to engineering or technology teams and supported them through delivery.
• You can explain to non-technical colleagues how a model or tool works, how well it performs and where its limits are.
• Будет плюсом: Data science experience in pricing, claims, exposure management or portfolio management, in insurance or another financial-risk setting; Knowledge of specialty or Lloyd's market insurance is helpful but not essential, as long as you are curious about how underwriting creates value and can learn the domain quickly; Training or fine-tuning models for text and document tasks when prompting alone is not enough; Experience as a product owner, or of working in agile teams alongside engineers and business users

• Условий в вакансии нет