How do you optimize a slow-running SQL query when working with millions of rows of patient-level data at Amgen?
