Skip to content

ML Sys Design


Notes on ML systems at scale — recommendation engines, feed ranking, retrieval architectures, and the engineering choices behind them.

Each piece is structured in two parts: the original source (a paper or engineering post), then where the field has moved since.


Notes

  • YouTube Recommendations — Deep neural networks for recommendation at scale. Based on the 2016 Google RecSys paper by Covington, Adams, and Sargin.

  • Facebook News Feed Ranking — Multi-objective ranking on a social graph. Based on Meta's 2021 engineering post on Feed ranking.