The API has recovered and is operating normally. The issue was caused by API pods restarting due to an unhandled edge case in jobs processing, which briefly made the system unresponsive. We're monitoring to confirm stability and will follow up with a root-cause fix to prevent recurrence.