Integrating High Performance Feature Stores with KServe Model Serving

Explore the integration of high-performance feature stores with KServe model serving in this conference talk by IBM experts Ted Chang and Chin Huang. Delve into the background of feature stores and feature engineering, understanding their differences and importance in machine learning workflows. Learn about Feast, a popular open-source feature store, and its three main concepts. Discover the challenges faced with KServe and how they are addressed through various deployment scenarios. Gain insights into the KServe overview and Model Mesh architecture, followed by a scalability test highlighting the system's performance. Witness a practical demonstration of making predictions and setting up a cluster. By the end of this talk, grasp the key concepts and practical applications of integrating feature stores with KServe for efficient model serving in production environments.

Introduction
Agenda
Background
Feature Stores vs Feature Engineering
Feast
Three Main Concepts
Problems with KServe
Deployment Scenario
KServe Overview
Model Mesh Architecture
Scalability Test
Highlights
Making a Prediction
Demo
Cluster setup
Recap