What are the trade-offs between using Python and Scala for building large-scale data transformation pipelines?
