The Lab
Experiments in quantifying the uninsurable.
I am currently building AION, an AI-native underwriting engine designed to bridge the gap between technical space data and insurance capital.
This feed documents the build in real-time: the technical bottlenecks, the architectural pivots, and the solutions required to deploy a working Risk Engine.
Active Module: M2 (Context & Explainability)
AION: Module 1 Phase 3-5
By the end of Phase 2, AION's risk engine could model realistic failure curves and price missions—but it wasn't production-ready. Phases 3, 4, and 5 transformed it from prototype to platform: 7 phase-specific actuarial models, regulatory compliance tracking, Lloyd's-grade UI, and <0.3s response times. Built using AI agent orchestration across research, planning, and execution. The result: a dependable underwriting workbench that real underwriters can trust. This is the story of that transformation—the obstacles, the iterations, and the systems thinking that made it possible.
AION: Module 1 Phase 2
Phase 2 of AION’s Risk Engine was about moving from a working prototype to something an underwriter could actually trust. I rebuilt the core logic around survival curves, calibration, environmental sensitivities, and a clearer mission view to make risk reasoning easier, not flashier.