Tell me about a time when you discovered a critical production bug shortly before or after a release. How did you communicate the issue, what steps did you take to mitigate impact, and what changes did you implement to prevent a recurrence?

Shortly after a rollout I noticed error rates spike to 12% and support flagged 1,200 users seeing wrong account summaries. I immediately alerted the on-call lead and PM, posted a concise incident note to Slack and paged the SRE team. We toggled the feature flag to stop the rollout within 18 minutes, which reduced errors to baseline. I led a small team of six to reproduce the issue, pinpointing a pagination offset bug that caused cross-tenant data leakage under certain sort parameters. We shipped a hotfix in two hours and coordinated a customer notification. Postmortem actions included an automated regression test covering the bug scenario, a new canary rollout policy, and a runbook; those changes cut similar incidents and reduced mean time to recovery from 90 to 25 minutes over the next quarter.