twitter/the-algorithm ? reverse-engineered prompt

I'd like to build a social media recommendation engine, inspired by the open-source X algorithm. The main goal is to create a personalized "For You" timeline that shows users a feed of posts. This feed should be a mix of content from people the user already follows (in-network) and recommended content from accounts they don't follow (out-of-network).

The system needs to rank all these posts to figure out what's most relevant. It should use signals like a user's past likes, follows, and interactions to make good predictions. I also want it to calculate some kind of reputation score for users and filter out unsafe or NSFW content.

The repository looks like a big collection of microservices in Scala and other languages. You'll have to figure out how all the different parts, like the "home-mixer" and the various ranking models, work together to produce the final feed. The build instructions seem incomplete, so you'll need to set up the Bazel files to get everything compiling and running.