At scale, large solutions are sometimes required to tackle even the smallest tasks, and ML is no different. Comcast is building architectures to handle end-to-end ML pipelining and deployments.
Nick Pinckernell outlines a solution that demonstrates configuration-based, continuously integrated and deployed solutions to handle data transformation, normalization, and model serving. This is accomplished using a range of tools and frameworks such as Kubernetes, Apache Spark, and more. It all starts with a large Apache Spark environment used by many researchers to explore and train models. The researchers are then empowered to develop simple or complex model graphs and deploy themselves using Kubeflow and Seldon Core. Data streams into these models using Apache Kafka with windowing and aggregation handled by Redis.
You’ll gain an understanding of the architecture, configuration, and technologies that are involved in making this happen at scale. Nick provides specific examples and flows of requests to example models to demonstrate all the necessary components and configuration.
Nick Pinckernell is a senior research engineer for the applied AI research team at Comcast, where he works on ML platforms for model serving and feature pipelining. He’s focused on software development, big data, distributed computing, and research in telecommunications for many years. He’s pursuing his MS in computer science at the University of Illinois at Urbana-Champaign, and when free, he enjoys IoT.
©2019, O'Reilly Media, Inc. • (800) 889-8969 or (707) 827-7019 • Monday-Friday 7:30am-5pm PT • All trademarks and registered trademarks appearing on oreilly.com are the property of their respective owners. • firstname.lastname@example.org