Design MLOps pipelines for continuous ML deployment