Perilyx AI is headquartered in Hartford, Connecticut — a deliberate choice. We work down the street from the reinsurers and carriers who are our first customers, not a time zone away from them.
Perilyx AI was founded in 2024 after our team spent years watching catastrophe models get the first event roughly right and the second event completely wrong. A wildfire model would correctly flag a burn area, then miss the debris-flow claim eighteen months later entirely — because nothing in the model tracked how the burn had changed the watershed.
We built Perilyx to close that gap: a simulation engine that treats climate hazards as a connected chain of physical state changes, not a set of independent line items.
Former catastrophe modeling lead at a major reinsurer. PhD in atmospheric science. Led the team that first flagged systematic underpricing of compound wildfire-flood sequences.
Built large-scale physics simulation infrastructure prior to Perilyx. Leads the neural PDE solver research that underpins the cascade engine.
Two decades in reinsurance underwriting leadership. Translates Perilyx's model output into workflows underwriting teams actually use.
Open roles span climate ML research, simulation infrastructure, and insurance partnerships — all based at our Hartford headquarters with flexibility for hybrid work.
Neural PDE modeling for hazard propagation. PhD in climate science, physics, or ML preferred.
ApplyScale our solver infrastructure across multi-region GPU clusters processing satellite-resolution data.
ApplySit between our science team and underwriting customers, translating cascade scores into workflow integrations.
Apply