Session language – English Target audience – Developers, Data Scientists, R&D

It’s good for feature reuse in machine learning, thereby increasing data science accuracy and velocity.

A feature store is a single interface to create, discover, and access features for model training and inference. A holistic feature store solution should be capable of:

• Ingestion - both from streams and batch jobs
• Serving - low latency single features for inference and high throughput bulk features for training
• Feature Engineering - transforming and aggregating
• Discovering - features and how to retrieve them

This session will attempt to demonstrate why a feature store is useful, review current open source solutions, and suggest how to build one.