Ooru: three-word addresses and a citizen power-cut map for Chennai
A three-word address for every 12.5 m of Chennai, and a live map of power cuts that resists abuse without asking anyone to log in.
- What
- My own product · M.Tech GIS project, SRM
- Role
- Designer and engineer
- When
- Aug – Sep 2026
- finest cell, in a nested 5 km → 250 m → 12.5 m grid
- 12.5 m
- to cluster 1,900 active cells into outages
- 7 ms
- anti-abuse by network identity, corroboration and Turnstile
- 0 logins
The situation
Street addresses in Chennai are hard to give and hard to find. Ooru gives every 12.5 m square three words. On top of the grid sits a citizen-reported power-cut map, which has to stay honest without asking anyone to sign up.
What I built
- Nested three-word addressing on a UTM grid (5 km → 250 m → 12.5 m) from a 936-word lexicon, placed by an ML pipeline: MiniLM embeddings, a self-organising map, then Hungarian assignment.
- /outages: citizens report power cuts on a live map. Anti-abuse works without logins: network-hashed identity, corroboration from three networks, per-ISP witness counting, rate limits and invisible Turnstile. It deliberately stores no history.
- Next.js on Cloudflare Workers with D1, rendered with deck.gl and MapLibre; outages cluster in 7 ms at 1,900 active cells.
Stack
- Next.js
- Cloudflare Workers
- D1
- deck.gl
- MapLibre
- Turnstile
- Python (embeddings, SOM)
