You'll be surprised at what "weak" code you'll see in the real world and have to explain to someone it's not maintainable or clean or whatever.

This is especially true of junior data scientists (in my anecdotal experience so far)