About Dwell
Dwell is a real-time transit platform built for university campuses. It turns the raw GPS coming off campus buses into arrival predictions riders can actually trust, and proves their accuracy in the open.
Why it exists
On a lot of campuses, the times riders see don't match the bus that actually shows up. Dwell takes a different approach: rebuild every prediction from the ground truth (where the vehicles really are) and measure the result continuously so the quality is never a mystery.
One platform, end-to-end
Dwell owns the whole lifecycle end-to-end, so each stage can make the next one better: cleaner data upstream means better predictions downstream, and every trip feeds the history and accuracy record that sharpen what comes next.
Raw GPS pulled in from on-bus devices and agency feeds.
A Python engine on Google Cloud snaps positions to routes and builds predictions.
Every observation lands in BigQuery to learn per-segment timing.
Standards-based GTFS-Realtime, served from the Cloudflare edge.
Native iOS (Swift) and Android (Kotlin) rider and driver apps.
Predictions scored against established benchmarks, published openly.
Built for the academic calendar
A campus in July moves nothing like a campus in October. Dwell learns separate timing models for each service period (summer, semester, and breaks), so arrivals reflect how the campus actually moves right now.
Who's behind Dwell
Dwell is designed, built, and operated end-to-end by its founder: one engineer responsible for everything from the GPS ingestion and the prediction engine to the native apps in riders' hands and the public accuracy report.