Worked on the derailed/cilium repository to improve observability reliability by addressing a critical bug affecting Hubble metrics exposure. Focused on ensuring that the ServiceMonitor for Hubble metrics is correctly created when dynamic metrics are enabled, the solution updated Helm chart conditions to account for both hubble.metrics.enabled and hubble.metrics.dynamic.enabled flags. This change allowed Prometheus to continue scraping metrics seamlessly in dynamic configurations, reducing manual intervention for operators. The work involved Go templating and Kubernetes resource management, resulting in a minimal and maintainable code footprint. Documentation and onboarding materials were also enhanced to support smoother adoption of dynamic metrics features.