How would you design a CI/CD pipeline using Docker, Kubernetes, and Vertex AI to ensure zero-downtime redeployment of a model for X Development?