How do you defend your past project decisions and explain why at every step in a data engineering context?