I know that's too much to ask, but if you implement simple into a "chooser" in https://github.com/avantoss/loop/tree/master/lib/choosers you'll have at least one company using this method!

We use a solution inspired by Jasper Snoek's Spearmint, which allows to somewhat parallelize bayesian search through a concept called "fantasies". If you've never read his paper I encourage you to find it on arxiv. The most gentle and thorough introduction to this topic that I've seen.