Netflix uses LLM to evaluate the quality of show synopses, combining expert and user feedback.
Netflix develops interval-aware caching for Druid to reduce the load of repeated queries.
AWS solution combines CV and generative AI to automate workplace safety monitoring.
Meituan BI builds a metric platform to address issues of confusing data definitions and poor query performance in traditional BI under personalized datasets.
Zhiyuan.com practices Flink on K8s, optimizing resource and operational efficiency.
Kuaishou's advertising platform migrated from a ClickHouse+ES architecture to Apache Doris.
Yelp's production-grade AI assistant uses a RAG architecture to overcome challenges.
LinkedIn replaced ZooKeeper with Kafka and xDS to restructure its service discovery system for large-scale operations.
Kuaishou reveals its AI development paradigm, highlighting that individual efficiency improvements do not equate to organizational efficiency gains.
Netflix uses the Data Canary system to validate catalog metadata.
Fz AI Testing achieves significant optimization in test case maintenance, missed testing, and dead loops.
DeXu uses Apache Doris to replace Greenplum, achieving performance gains and cost reductions.
Xiaomi Doris Lakehouse Practice, Unified OLAP, Storage and Compute Separation
Meituan shares AI programming quality assurance practices.
Kuaishou OneRec rebuilds the recommendation system with large models, reducing costs by 1/10.
Yingjiao Network migrates to StarRocks to address ClickHouse bottlenecks.
Spotify built its own data platform, processing 1.4 trillion data points daily.
Dassault and Zhongnan Institute launch ZDPLM platform for intelligent lifecycle management in construction
Apple releases a 400,000-image AI image editing dataset.
Duolingo shares FinOps practices, transforming cloud expenditure into actionable engineering insights.