Case Studiesintermediate12 min · 8 steps
Recommendation Engine with Explanations
Rank 50M users' movies with classic ML in under 200 ms, and spend the LLM only on cached, fact-checked explanations.
requestheavy workdataresponsetoken
metrics are illustrativeProblemStep 1 of 8
Recommend a movie — and say why
A streaming platform with 50 million users wants recommendations that come with a natural-language reason: “Because you enjoyed Inception, you might like Tenet for its mind-bending time mechanics.”
5M daily actives each load 10+ recommendation sets a day. Every set must render within a 200 ms p95, handle brand-new users, and never leak one person's viewing history to another.
InsightTwo jobs hide in one feature: ranking (fast, huge volume) and explaining (language, personal). They don't need the same tool.
Lesson 11 of 20 in Case Studies