Pymoo

If you're optimizing anything with multiple conflicting objectives, this is the framework you want. It implements the heavy hitters like NSGA-II and NSGA-III for finding Pareto fronts, handles constrained problems cleanly, and includes benchmark suites like ZDT and DTLZ for testing. The interface is consistent across algorithms, which matters when you're experimenting. It scales from simple single-objective GAs to many-objective problems with 5+ objectives. The constraint handling is thoughtful, defaulting to feasibility-first but giving you penalty methods when needed. Honestly, the main value is that someone already implemented these algorithms correctly so you don't have to debug your own NSGA-II implementation at 2am.