During February 2026, Andrianta focused on stabilizing Azure Service Bus messaging within the scaleapi/llm-engine repository, addressing persistent idle-timeout and 503 errors that impacted backend reliability. Leveraging Python and Azure services, Andrianta implemented proactive keep-alive heartbeats, robust retry policies, and startup reconnect logic to ensure consistent message delivery. The work included upgrading the azure-servicebus dependency to resolve idle-connection regressions and removing unreliable broker connection pooling. By exposing environment-driven configuration for keep-alive intervals, Andrianta enabled safer operational tuning. This targeted backend development improved the resilience of API integrations, demonstrating depth in diagnosing and resolving complex distributed system issues.