Haven — Environments That Adapt to People

The concept

Most "smart" spaces make people adapt to them: apps, remotes, routines to configure. Haven flips that. The space senses context and adjusts itself, quietly, toward comfort and safety.

The hard part

This is a systems problem before it is an ML problem: what to sense, where inference should live (edge vs hub), and how to keep behavior legible so a room never feels like it is acting on you without reason. I see it as the long-horizon home for the sensing and control work I am doing now.