Cirium

AWS flight-data pipelines for aviation analytics. On-call checks from an hour to three minutes, 90 or more vulnerabilities cleared.

Role

Engineer 2

Period

2022 to 2024

Stack

AWS Lambda · Node · Python · Docker · Redis

Outcome

213M+ events a day

Two years on mission-critical aviation analytics, running flight-data pipelines that carried more than 213 million events a day.

Where the standards came from

A silent failure in aviation data is somebody’s flight. That is not a metaphor about uptime — it is the reason the monitoring, the alerting and the security posture were treated as first-class work rather than cleanup.

What changed

  • On-call checks from an hour to three minutes. The daily verification routine was manual and nobody wanted to run it at 2am, which meant it sometimes did not get run. Automating it saved an estimated $45K a year and, more usefully, made the check actually happen.
  • 90 or more security findings cleared across the pipeline estate.

This is where I learned to care about the checklist nobody wants to run — and to build the thing that runs it for them.

← All work