The context
CIBEST equips public-transport vehicles: cameras, GPS, passenger-counting cells. On board, the Scene5 recorder; on the operator’s side, the Supervision web app. I did my final Master’s internship there, and the team kept me.
What I delivered
- Counting cells integrated into the Scene5: configuration, real time over RabbitMQ from the C++ layer, propagation to Supervision.
- Ridership map in Supervision (MapLibre), filters, AVL ↔ GTFS matching, CSV export.
- Automatic PostgreSQL archiving with hot/cold partitioning: the database manages its own retention.
- Grafana + k6: performance measured, regressions caught, bottlenecks fixed.
Stack
Vue 3 + Tailwind, Bun + Hono + Prisma, PostgreSQL, Node.js and C++ on board, RabbitMQ, SQLite, Ansible and Yocto. The same family of tools as LamaQuest — productive from week one.