This article explains the evolution of Vivo's real‑time recommendation feature‑concatenation architecture, compares hour‑level, Redis‑streaming and RocksDB state‑backend solutions, and details the memory, performance, startup and HDFS RPC problems encountered along with the concrete fixes applied.
This article dissects the architecture of a modern short‑video backend, covering layered system design, core services such as video production, distribution, interaction, storage strategies, real‑time and offline recommendation engines, high‑concurrency streaming solutions, and practical techniques for cost control, scalability, and fault tolerance.
This article presents Kuaishou's graph database storage‑compute separation architecture, detailing its application in real‑time recommendation scenarios, core requirements of cost, performance and usability, the layered service design, memory‑compact models, edge structures, snapshot isolation, and key performance optimizations such as Share‑Nothing and columnar data flow.
This article presents the design and implementation of the Real‑Time User Understanding Service (RTUS) for the Fliggy travel platform, detailing its architecture, multi‑chain data fusion, model and data reuse techniques, and several AI‑driven algorithms for cold‑start interest representation, intent prediction, and destination forecasting, together with extensive offline and online experimental results.
This article details Meituan's interactive recommendation system for its food‑delivery homepage feed, covering the motivation, challenges, system architecture, user intent modeling, evaluation metrics, experimental results, and future directions, illustrating how real‑time, user‑centric recommendations improve conversion and user experience.
This article details the WeChat engineering team’s practical experience deploying and optimizing Apache Pulsar on Kubernetes for massive real‑time recommendation workloads, covering cloud‑native advantages, non‑persistent topics, load‑balancing tweaks, broker cache improvements, COS offloader development, and future roadmap.
This article details the architecture, incremental model training, feature engineering, and deployment strategies that enabled NetEase Cloud Music to achieve real‑time live‑stream recommendation, covering business background, multi‑objective modeling, real‑time feature pipelines, sample attribution, feature admission, and online performance results.
This article explains how Alibaba’s local‑life platforms leverage edge‑side AI to run machine‑learning inference on users’ devices, detailing the concept, advantages, technical architecture, and concrete implementations such as user feature extraction, intelligent recommendation, and smart push, while outlining future directions.
This article presents a practical study on modeling real‑time positive and negative feedback sequences and applying multi‑objective optimization in the re‑ranking stage of Taobao Live, detailing system architecture, feature engineering, loss design, experimental results, and future research directions.
This article details ByteGraph, the high‑performance graph database behind TikTok, covering its motivation over relational stores, data model, Gremlin query language, multi‑layer architecture, indexing, hot‑spot handling, and offline‑online data pipelines, while answering common technical questions.
This article describes how Tubi’s engineering team built and optimized a real‑time recommendation backend, using ScalaMeter microbenchmarks and wrk2 load testing to measure latency, throughput and error rates, and demonstrates scaling the service across multiple machines with custom scripts.
This article explains how to use the open‑source Kiji framework together with HBase, Avro, and MapReduce to build a scalable, entity‑centric real‑time recommendation system that can instantly refresh suggestions based on user context and recent interactions.
