How did you ensure data quality and accuracy in the pipelines you built, particularly when handling large volumes of data?