After seeing DeepMind's AlphaEvolve [1], I wanted to try building something similar. This isn't the first open-source attempt [2] [3] [4], and it won't be the last, but I focused on making mine more modular and customizable to make experimenting with different evolution-based approaches or prompting strategies easier (e.g. balancing exploration and exploitation).

The vision is to make it easy to swap components and experiment, though there's still work to do.

I'd love any feedback.

GitHub: https://github.com/larose/optiverse

[1] https://news.ycombinator.com/item?id=43985489

[2] https://github.com/codelion/openevolve

[3] https://www.saipraneeth.in/ml/building-evolve

[4] https://github.com/paperswithbacktest/pwb-alphaevolve