Case Study
Implemented a low-latency, dynamic pricing calculation layer integrating machine learning risk models with classical insurance rating grids for instant, automated underwriting decisions.
Traditional insurance pricing relies on rigid rating grids maintained in spreadsheets or legacy databases. When risk landscapes change rapidly, these static grids fail. I built a modern, microservice-based dynamic pricing calculation layer that runs machine learning-driven loss predictions directly alongside actuarial formulas.
The engine exposes a high-throughput, sub-100ms JSON API. It ingests historical loss runs, weather risks, real-time macro-economic factors, and demographic trends to calculate a dynamic risk score. This score then scales the baseline premium calculated from traditional insurance rating books, optimizing both premium yields and exposure risk.
This system allowed Insly's clients to deploy new, risk-adjusted pricing algorithms into the market in days rather than months. By automating complex rating decisions, the dynamic pricing engine increased auto-binding rates from 22% to over 68% while keeping loss ratios strictly bounded.