That has been a huge problem in the past and it's still somewhat of a problem. However now that pip and virtualenv (renamed venv) are part of the standard python install and pip has gotten a lot better when comes to cross platform installs and handling binary dependencies, hopefully we'll see some more standardization around those tools.

That being said, I use conda for managing dependencies and environment for basically all my python development work and have no complaints.

pip: install a package system-wide

venv: create fake system paths locally in your current directory

so do everything within a venv and pin your dependency versions into requirements.txt. i've found it to be exactly as powerful as i need, works out of the box, and it's easy.