Coming Soon! … Read More
High Performance Feature Serving with Feast on AWS
In this lightning talk we will showcase how teams can deploy and productionize a feature store on AWS with Feast. We will demonstrate how users can train a machine learning model using historical features from Redshift, deploy the model into production on AWS, then serve fresh feature values to the model using a serverless feature serving stack with DynamoDB. … Read More
How Shopify Contributed to Scale Feast
This talk will discuss how Shopify manages large volumes of ML data (billions of rows) using Feast. Shopify decided to adopt Feast to build their ML Feature Store in early 2021. We will speak about how we contributed to Feast to make it more scalable (example of PRs we have contributed at https://github.com/feast-dev/feast/pull/1602) … Read More
Panel: Building High-Performance ML Teams
As Machine Learning moves to production, ML teams have to evolve into high-performing engineering teams. Data science is still a central role, but no longer sufficient. We now need new functions (e.g. MLOps Engineers) and new processes to bridge the gap between traditional data science and the world of software engineering. In this panel discussion, we’ll discuss how high-performing ML teams are organized to build and deploy production-quality ML models with engineering best practices. … Read More
How Robinhood Built a Feature Store Using Feast
Features are essential to ML models. Therefore, a good feature infrastructure is important to any organization that wants to use ML properly in production. Feast is a great tool for building up your feature infrastructure. However, using Feast in production may need customization of your tech stack, extension for advanced use cases, and improvement of reliability and observability. In this talk, we will share the lessons learned from how Robinhood built a feature store from Feast. … Read More
