Caderneta: OCR-powered bet tracker
A multi-user web app that reads screenshots and turns them into data and stats.
- React
- OCR
- Docker
- nginx
- Jenkins
Problem
Logging bets by hand is slow and error-prone, and the data ends up scattered across screenshots.
Solution
A React frontend with login and a backend that extracts the data from screenshots with OCR and stores only the result (images never stay on the server). A shared page shows pre-match analyses, predictions and hit rate. It runs in Docker behind nginx on its own Cloudflare subdomain, with deployment moving to Jenkins.
Result
In production since October 2026, with real data and several users.