Real-time ML enables new classes of systems that can quickly adapt to environments, such as real-time fraud detection, intrusion detection, and real-time personalized-recommendation systems. Feature engineering is the most challenging aspect of for real-time ML, and online feature pipelines are the key component of any solution.

In this webinar, we will introduce the challenges of building online feature pipelines, including integration with a feature store (to avoid training/serving skew), online model serving infrastructure, and real-time feature computation, with stream processing and on-demand features. We will present these challenges in the context of the Hopsworks platform, and show how it reduces the time required to put real-time ML in production.