convexengineering/gpkit ? reverse-engineered prompt

Build me a Python package for geometric programming that lets me define optimization models in a natural way and then solve them with a backend solver without me having to worry about the solver details.

I want it to feel usable for engineers, with clear examples and basic documentation, and it should support the kinds of models you’d expect for geometric programming. Please make sure it can run unit tests and includes installable package setup so I can use it locally. If you need to check current solver docs or best practices online, go ahead and do that.