In 2026 I built real-time API alerting on Cloudflare, solo and end to end. It watches the critical payment, auth and KYC journeys and flags problems before customers report them.
Context
Problems on critical customer journeys needed to be detected before customers noticed and reported them.
My role
I built it alone, end to end.
Approach & trade-offs
I evaluated several options. A heavy ELK stack was one of them, but no log storage was needed, so I chose a lightweight query-based monitor instead.
- Coverage: the critical payment, auth and KYC journeys.
- Per-endpoint thresholds, so each journey is judged on its own normal behaviour.
- Alert dedup, so one problem raises one alert rather than a flood.
- A daily heartbeat, so silence means healthy, not broken.
Results
- Earlyissues detected before customers report them
- Solobuilt end to end
- Lightweightno log storage needed
It detects issues before customers report them, and it helped surface a credential-stuffing attempt early.
More case studies
- System Health Dashboard: A Self-Service Status Page for Support, the CTO and On-Call
- Affiliate Portal: Architecture, Deployment Security and CI/CD for an In-House Platform
Work with me: SRE, observability and alerting · Hire or work with me · My background