How would you design a scalable machine learning pipeline on a hybrid-cloud architecture for Cloudera customers?
