How do you implement checks in a pipeline to catch nulls, duplicates, or malformed data before it reaches downstream users at Barclays?
