Netflix replaces its 15-year-old recommendation algorithm with LLM-based ranker GenRec
VeryWellVersed · x · 2026-09-29
Netflix has replaced its decade-old recommendation engine with GenRec, an LLM-backed ranker.
- Old system: thousands of handcrafted features, bespoke architectures, massive infrastructure, constant manual tuning.
- New approach: watch history and metadata are converted into a natural-language sentence and fed to a foundation LLM, which reads user behavior like a story and scores the entire catalog in a single forward pass.
- No manual feature engineering required — a paradigm shift from feature engineering to semantic understanding for recommendations.
More from Companies & People
- Meta Opens Research Scientist Roles in Montreal for World Models, Video Gen, RL — _dsevero · 2026-09-29
- SVSD partners with Clay and DoorDash on student design contest, finalists flown to NYC — MxMnr · 2026-09-29
- Cognition Researchers to Lead a Session on Globally Distributed RL Training, Oct 5 — _shreya_s · 2026-09-29
- YC-Backed Invertix Deploys AI Agent Swarms to Run Energy Assets, Eyeing Data Centers — ycombinator · 2026-09-29
- AI productivity dashboards look green, but enterprise decision quality isn't improving — DavidLinthicum · 2026-09-29
- MIT's Phillip Isola: counterintuitively, now is a great time to do an AI PhD — phillip_isola · 2026-09-29