Eh, SQL and python can still work together very well where SQL takes the place of pandas. Doing things in waves/batch helps.

Big problem with pandas is that you still have to load the dataframe into memory to work with it. My data's too big for that and postgres makes that problem go away almost entirely.