Describe a scenario where you had to optimize a slow-running Python ETL pipeline. What profiling tools and techniques did you use?
