Back to portfolio
ObservabilitySenior EngineerObservabilityJUN 2024 — PRESENT

Incident Intelligence

Correlation tooling that groups alerts, deploys, and dependency signals into a single incident narrative for responders.

GoPrometheusClickHouse

Responders rarely lack data during an incident. They lack a narrative that connects the alert to the change that caused it.

Problem

A single failure produced dozens of pages across teams, with no shared view of what changed and when.

Approach

  • Grouped alerts by dependency topology rather than by label equality
  • Joined the alert window against deploy and feature-flag events
  • Published one incident timeline that every responder could read
  • Kept the raw signals one click away for anyone who distrusted the summary

Result

Coordination overhead dropped noticeably, and postmortems started from a reconstructed timeline instead of a manual reconstruction from chat history.