Airbus Defence and Space
Test automation and observability for critical internal applications.
Context: Briefing
At Airbus Defence and Space, a failure that nobody notices is the expensive kind. I joined an engineering team in Élancourt to make the testing and monitoring of critical internal applications automatic, observable and trustworthy.
The work spanned the whole loop: tests that run on their own, metrics and alerts that say what broke, a dashboard to drive it, and the documentation that lets the team keep operating it.
What I did: Objectives cleared
- Automated 40+ end-to-end test suites and 4 production monitoring scenarios in Python with Robot Framework, Selenium and Appium, across web and Android applications.
- Built the observability and alerting stack (Prometheus, Pushgateway, Grafana, Alertmanager, Statuspage), from metric collection to alert routing and incident escalation.
- Containerized the platform with Docker Compose behind an Nginx reverse proxy, delivered as a reproducible, versioned deployment.
- Wrote distributed Python services (Celery, Redis) that run test workloads asynchronously and in parallel across worker pools.
- Designed a real-time orchestration dashboard (React, TypeScript, Vite, Tailwind, REST APIs) to schedule, trigger and follow automated pipelines.
- Automated recurring Linux operations in Python and Bash, and wrote the runbooks.
Hardest problem: Boss fight
Alerts that say what broke, not just that something did
A red dashboard is useless if nobody can act on it. I pushed test results into Prometheus through Pushgateway, routed alerts with Alertmanager to the right people and to Statuspage, and traced each recurring failure down to the faulty module in code I had not written.
Results: Rewards
- ~3 h of manual QA removed per engineer, per day
- From prototype to daily production use
- Faster releases with test coverage on critical applications