Pip gives you whatever the default build is, but for numpy they build optimized builds into wheels for almost every platform. https://pypi.org/project/numpy/#files For anything else, you have to get into the details of how it is built and what the options are which is going to depend on the specific package. Building numpy yourself is not beginner friendly, you have to know a ton of details about linear algebra packages.

Conda is helpful for the odd cases where one C dependency conflicts with another and you need to be careful specifying which build you want.