Building Sentinel in Two Weeks
On March 30, 2026, I lost my job at EA. By April 8, I had built the largest open surveillance awareness map in the world: 417,606 cameras across 103 countries on 6 continents.
Here’s how.
Week 1: Foundation
Day 1-2: Incorporated in BC, acquired symvek.com, set up infrastructure. Built Shield (browser privacy extension) to feature-complete status.
Day 3-4: Built the data pipeline. Python ingest scripts for EFF Atlas, OpenStreetMap, FLOCK dataset, and government open data APIs. The first dataset had 354,000 cameras, mostly US.
Day 5: Deep research on counter-surveillance market. Discovered no product combines surveillance mapping, detection, and defense. The market gap is real.
Day 6-7: Launched the web map. MapLibre GL + PMTiles. Solved the Cloudflare Workers Range header issue by pre-fetching the entire file into an ArrayBuffer client-side. 354K cameras live on production.
Week 2: Global expansion
Day 8-10: Built ingest scripts for UK (TfL), Canada (DriveBC, 511 networks), Europe (15 countries), Asia-Pacific, and Latin America. Pushed from 354K to 417K cameras across 103 countries in 48 hours.
Day 11-12: Built Sentinel Mobile app (8 screens, 9 services), Sentinel Intelligence API, counter-drone detection prototype. Wrote 50 blog posts. Fixed UX issues. Added Firefox support for Shield.
The tools
- Astro for the website (SSR on Cloudflare Workers)
- MapLibre GL JS for the web map
- PMTiles for efficient tile storage
- tippecanoe for tile generation
- Expo/React Native for the mobile app
- Cloudflare Workers for deployment
- Python for data pipeline
- Claude Code for AI-assisted development
The key insight
The surveillance data already exists in public records. Nobody had aggregated it into a single, queryable, consumer-friendly product. The engineering isn’t revolutionary. The aggregation is.
417,606 cameras. 103 countries. 50 blog posts. Two weeks.