How would you design a scalable pipeline to continuously train and redeploy models as new clinical data becomes available at GSK?
