Developed and integrated a new runtime capability for the opendatahub-io/kserve repository, focusing on expanding model serving options with minimal operational complexity. The work centered on adding vLLM as a supported runtime, enabling both classification and text embedding services within KServe. This involved updating configuration files, defining runtime specifications, and creating testing scripts to ensure robust integration. The implementation emphasized maintainability and traceability, linking all changes to specific commits and validating the solution through comprehensive end-to-end tests and CI checks. Utilized Python, YAML, and Kubernetes, applying CI/CD practices to deliver a reliable, extensible foundation for future runtime integrations.