Worked on the red-hat-data-services/odh-model-controller repository to update the VLLM API server startup process, focusing on aligning it with newer VLLM versions and configurations. The approach involved modifying the YAML-based configuration to switch the executed Python module from vllm_tgis_adapter to vllm.entrypoints.openai.api_server, which improved compatibility with current VLLM deployments and reduced upgrade risks. This change streamlined the startup flow, removed legacy dependencies, and simplified future maintenance. Emphasizing configuration management and DevOps practices, the work ensured production readiness and traceability through clear documentation and commit history, supporting more reliable and maintainable deployment workflows for the project.