Having architected some large projects, then coded, then refactored because original designs didnt work out, eventually getting a working project that does thin

Having architected some large projects, then coded, then refactored because original designs didnt work out, eventually getting a working project that does things. Also now having done the same thing 95% ai coding. Good taste is good architecting. Knowing what to use; how to use it and when. How to maximize KISS. Simplicity is key. When is it right to keep it very simple and load it to a dict? When to upgrade to a panda/numpy dataframe? When to go sqlite3? When to go full on DB(postgresql?). Do you need to go that extra step in order to stay with your current design; will the code become complicated and ugly to do that? What are the upsides and downsides? good taste is the comfort to try a new library that looks like it'll do a better job than what you know how to use now. Does your script run in memory for long? Is that ok? Recently I learnt about Python Slots and use them now. You look at: https://www.w3schools.com/python/python_datatypes.asp https://www.w3schools.com/python/python_datatypes.asp And for some reason they dont even mention slots. but omg so much better!