Building Real-Time ML Features with a Feature Platform

Explore the challenges and solutions for deploying machine learning in production with a focus on real-time feature engineering. Learn how feature stores and feature platforms have emerged as essential tools for scaling ML efforts from industry experts Mike del Balso and Willem Pienaar. Discover the unique requirements of production ML pipelines, including processing historical and fresh data, ensuring training/serving parity, and maintaining point-in-time correctness. Gain insights into the experiences of building core ML infrastructure at Uber and Gojek, and understand how feature stores enabled these companies to scale their ML efforts to thousands of production models. Delve into the evolution of feature stores to comprehensive feature platforms that manage the entire lifecycle of real-time ML features, including automated data pipelines for transforming raw data from batch and real-time sources. Acquire valuable knowledge on overcoming data challenges in production ML and streamlining project timelines for more efficient ML deployment.

Building Real Time ML Features with a Feature Platform