Building a Cloud Native Feature Store with Feast on Kubeflow
Talkat KubeCon + CloudNativeCon North America 2020 Virtual · on Friday, 2020-11-20· at 17:55· Intrado Virtual Event Platform,·
summary
Feast is a cloud-native feature store that deploys via Helm to Kubernetes, uses Postgres for a central registry, runs Spark ingestion jobs for batch and offline-to-online, and serves low-latency features from Redis for real-time prediction. It enforces point-in-time correctness for training vs serving, supports Kafka streaming and Parquet batch sources, and demonstrates end-to-end ingestion and online lookup working in the demo. The project reports 43 contributors and 6+ enterprise customers, is under the Linux Foundation, and partners with Tecton.
AI-written summary · about
chapters (6)
- 0:00Introduction to Feast
- 0:36Common ML data challenges
- 3:42Feast architecture and role
- 7:40Deployment and usage demo
- 26:46Project vision and growth
- 30:55Roadmap and conclusion
related talks
31mFeast as a Cloud Native Data System: Bridging Batch Processing and Online Serving...
36mIntegrating Feast Online Feature Store with KFServing
27mFeast: feature store for Machine Learning
33mFeathr, an Open Source, Battletested, Scalable and Enterprise-Grade Feature Store
23mFeathr: An Enterprise-Grade, High Performance Feature Store
51mSelf-serve Feature Engineering Platform Using Flyte and Feast
08mFeature Store Usage Patterns - From a Single Data Scientist to an Enterprise
46mScaling ML Pipelines with Feast, Ray and Kubeflow