https://mechanicalsnail.com/series/recommendation-systems/Recent content in Recommendation Systems on Mechanical Snailhttps://mechanicalsnail.com/images/logo.pnghttps://mechanicalsnail.com/images/logo.pngHugo -- 0.152.2en-usWed, 10 Jul 2024 14:00:00 +0100https://mechanicalsnail.com/posts/recommendation-systems-part6/Wed, 10 Jul 2024 14:00:00 +0100https://mechanicalsnail.com/posts/recommendation-systems-part6/Part 6 of 6: foundation models, LLMs and multimodal understanding in recommendation — what changes once the ranker can actually read the content it is ranking.https://mechanicalsnail.com/posts/recommendation-systems-part5/Wed, 10 Jul 2024 13:00:00 +0100https://mechanicalsnail.com/posts/recommendation-systems-part5/Part 5 of 6: a reference feed-ranking architecture — request flow, latency budgets, ads blending, two-tower retrieval and ranking implementations, and production monitoring.https://mechanicalsnail.com/posts/recommendation-systems-part4/Wed, 10 Jul 2024 12:00:00 +0100https://mechanicalsnail.com/posts/recommendation-systems-part4/Part 4 of 6: amplification harms, filter bubbles, fairness metrics and the governance controls that keep large-scale recommendation systems accountable.https://mechanicalsnail.com/posts/recommendation-systems-part3/Wed, 10 Jul 2024 11:00:00 +0100https://mechanicalsnail.com/posts/recommendation-systems-part3/Part 3 of 6: running recommenders in production — offline ranking metrics, A/B testing statistics, distributed training, model serving and compression, cold start, and drift.https://mechanicalsnail.com/posts/recommendation-systems-part2/Wed, 10 Jul 2024 10:00:00 +0100https://mechanicalsnail.com/posts/recommendation-systems-part2/Part 2 of 6: ranking and re-ranking — pointwise, pairwise and listwise losses, Deep & Cross and Wide & Deep architectures, multi-task learning, diversity via MMR and DPP, and bandits.https://mechanicalsnail.com/posts/recommendation-systems-part1/Wed, 10 Jul 2024 09:00:00 +0100https://mechanicalsnail.com/posts/recommendation-systems-part1/Part 1 of 6: the multi-stage architecture of a production recommendation system — candidate retrieval, embedding-based nearest neighbour search, and feature engineering at scale.