This advice isn’t specific to Python, but something that engineers who have spent a lot of time writing non-production code in Python Notebooks, do make the mistake of. This is not specific to Python either but is a good practice that’s useful when you are working with data engineers and data scientists who excel at data analysis more than writing production-ready code. I have tried Python’s Alpine-based images ( python:alpine ) and for any non-trivial project, it is very hard to use it due to Debian’s glibc vs Alpine’s musl differences.