[lysecret]

Interesting post. On thing to note, this seems to be about "on request" ranking. E.g. googleing something and in 500ms you need the recommended content.

However, a lot of usecases are time insensitive rankings. Like recommending content on netflix, spotify etc. (spotifys discover weekly even has a one week! request time :D).

In which case you can just run your ranking and store the recs in your DB and its much much easier.

[troiskaer]

This is pretty much what both Netflix and Spotify do. I would argue that there isn't a canonical recommendations stack that FAANG is converging towards, and that's a direct corollary of differing business requirements and organizational structure.