evaluation: which workflow works best for Python?

There are several ways to approach Python, so I wanted to compare experiences.

I am particularly interested in evaluation and collaboration. I would like to compare approaches that people have actually used rather than just theoretical suggestions.

Which approach would you recommend and why?