Detecting Anomalies in CI/CD Pipelines with ML

CI/CD pipelines frequently fail due to structurally different reasons, such as flaky tests, leading to significant hours spent on manual triage. Engineers often waste time scrolling through logs to identify recurring issues that a simple pass/fail signal cannot explain. The objective is to determine if ML can detect when a run looks unusual before human intervention. This approach aims to replace tedious manual digging with rapid anomaly detection, improving efficiency across the entire engineering team.