Weighted lotteries are used when we need to randomly select k of n items with given probabilities - commonly used by Outbrain's recommendations system to provide diversity for our users. Once we identified that our naive implementation is a production bottleneck, we decided to solve the problem in a fun way, and turned it into a coding challenge, followed by releasing creative benchmarked solutions along with classic algorithms to the community

https://github.com/guyko/weighted-lottery/wiki