We watch it, not just wait for a call
Every backend cron job — incident escalation, stale-trip cleanup, data archival — writes a heartbeat we check automatically. If one goes silent or fails, that's caught before it becomes a support ticket. Live status is public at
transitguide.uaroundu.com/status.
GPS history is archived before it's ever deleted
Trip location data moves through three tiers — full resolution for a week, downsampled for three months, then archived to S3 before we ever purge it from the database. Nothing is deleted without a copy existing first.
The database is backed up nightly, off the box
A full database snapshot ships to S3 every night, independent of the application server. If the EC2 instance itself were lost, the data survives it.
Errors reach a person, not just a log file
Unhandled backend errors are captured centrally so a regression is visible immediately, not discovered when a customer reports it.