Feathr, an Open Source, Battletested, Scalable and Enterprise-Grade Feature Store
summary
Open-source feature store built from five years of LinkedIn production use, integrated with Azure and accepted into the Linux Foundation AI & Data sandbox. Defines features over Kafka, Delta Lake, Snowflake, etc., materializes to Redis or Cosmos DB, runs compute on Spark, and optimizes tens of billions of rows with bloom filters and join-plan optimization and salted joins. Claims reduce feature rollout from weeks to days, reports 50% faster than some in-house pipelines and powers hundreds of LinkedIn models.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:33What is Feathr
- 5:49Feature store problems
- 14:30Feathr solution & architecture
- 24:01Enterprise scalability & security
- 25:03Feathr’s impact
- 28:53Summary
- 29:57Q&A
related talks
23mFeathr: An Enterprise-Grade, High Performance Feature Store
32mBuilding a Cloud Native Feature Store with Feast on Kubeflow
08mFeature Store Usage Patterns - From a Single Data Scientist to an Enterprise
32mA feature Store for Machine Learning
15mFeature Stores: Powering Machine Learning Infrastructure with Apache Flink
31mBuilding an open-source ML feature store with Apache Flink
20mNavigating the Skies of Data: A Flight Through Feature Stores
30mHow Feature Stores Enable Operational ML