During February 2025, Akira Matsukine focused on improving the stability of the rudderlabs/rudder-server repository by addressing a critical runtime issue in self-hosted RudderStack deployments. Akira identified and resolved a PostgreSQL container error caused by insufficient shared memory under high load, which previously led to 'No space left on device' failures. By increasing the shared memory allocation from 64MB to 128MB in both docker-compose.yml and rudder-docker.yml, Akira enhanced reliability and reduced downtime for enterprise users. This work demonstrated practical expertise in DevOps, Docker, and YAML configuration, with a clear focus on incident-driven resilience and production system reliability.