CCreator Rise Agency·Oct 13How TikTok's For You Algorithm Works: What TikTok Has Actually SaidWhat TikTok has officially said about its For You recommendation system: signals, weighting, diversity rules and cold start, explained for developers.6 min read·creatorriseagency.hashnode.dev
IInformed Clearly·Sep 19Building News Recommendations That Respect Reader ConsentHow Informed Clearly designs useful news recommendations around consent, visible reasons, editorial judgement, diversity and deletion.4 min read·informedclearly.hashnode.dev
CChinmay Singh·Aug 16How Recommendation Systems Control Your Digital LifeHow Netflix, YouTube, Spotify, and Amazon use AI recommendation systems to shape what you watch, hear, and buy.19 min read·chinmayssingh.hashnode.dev1
CClementina Tom·Jun 4Why Recommendation Systems Fail in Education (And What I Built Instead)Standard recommenders violate 81% of curriculum prerequisites. I built PLRS to fix it using knowledge tracing, forgetting curves, and a DAG.10 min read·clementina-tom.hashnode.dev21M
FFahad Shah·Apr 8Real-Time Recommendation System in 2026: Architecture, Algorithms, and SQLBatch-computed recommendations miss the purchase window. In 2026, streaming SQL closes the gap between what a user just did and what they see next. Here is the complete architecture.16 min read·risingwave.com
FFahad Shah·Apr 7Semantic Product Recommendations with Vector Search and Streaming SQLKeyword matching misses synonyms. Category filtering misses cross-category relevance. Vector similarity finds items that are genuinely similar in meaning, and streaming SQL keeps those embeddings fresh.16 min read·risingwave.com
FFahad Shah·Apr 7Frequently Bought Together: Build Amazon-Style Recommendations with Streaming SQLThe 'Frequently Bought Together' widget drives a significant share of Amazon's cross-sell revenue. Here is how to build it with streaming SQL so it updates the moment purchases happen.12 min read·risingwave.com
FFahad Shah·Apr 7Real-Time Collaborative Filtering: Build User-Based and Item-Based Systems in SQLCollaborative filtering is traditionally a batch ML problem. With streaming materialized views, you can maintain item similarity and user cohort signals continuously in SQL -- no model training required.19 min read·risingwave.com
FFahad Shah·Apr 7How to Build a Real-Time Recommendation System with SQLBatch recommendation pipelines miss the purchase window. This is the complete SQL architecture for recommendations that update within seconds of each user action.17 min read·risingwave.com
YYuheng LIANG·Feb 3Taste Aligner (四) Recommendation Engine v1:双区推荐(CZ/EZ)从 0 到可验收这篇文章记录我在 Taste Aligner 项目中实现 Recommendation Service v1.3 的全过程: 为什么要做“舒适区 / 探索区”双区推荐? 为什么必须拆成 Recall→Rerank? 如何让推荐“可解释、可控、可验证、Agent 友好”? 以及:我遇到的 2 个最难 bug(CZ 跨城污染、tags=[] 导致 EZ 消失)是怎么一步步修掉的。 0. 背景:Recommendation 在 Taste Aligner 里负责什么? Taste Al...4 min read·latenighttavern.hashnode.dev