Netflix: Towards LLM-Native Recommendation

About this event

In this technical meetup, we will unpack Netflix’s GenRec: Towards LLM-Native Recommendation at Netflix (https://medium.com/netflix-techblog/genrec-towards-llm-native-recommendation-at-netflix-f20be6f643e3), a report on an LLM-backed recommendation ranker designed for large-scale personalization. The session examines how Netflix verbalizes member history, item metadata, and request context; post-trains a Netflix-adapted foundation model for ranking; constrains outputs to in-catalog titles; and designs inference around real serving-cost constraints. GenRec presents a disciplined production pattern: context engineering, catalog-aware scoring, reward-weighted alignment, and prefill-only serving. Netflix reports that GenRec improved on a mature production ranker in both offline evaluation and a large online experiment while using substantially fewer labeled examples and input signals. Slides for past meetups posted: Github (https://github.com/YanXuHappygela/LLM-reading-group/tree/main) Recordings posted at: YanAITalk (https://www.youtube.com/@yanaitalk/videos) Feel free to reach out if you want to present at upcoming meetups! Note: You must have a Zoom account to login (free account is sufficient). Zoom Link will be posted to the event page one day before the meetup.