How often do you get the chance to learn from the leading references in a field, gathered in a single day and at no cost? That is what the Feature Store Summit 2026 offers on Tuesday, October 6.
This invitation is a personal recommendation. The summit is a free online conference, organized by Hopsworks, that brings together engineers who build the machine learning systems at Uber, Spotify, and Airbnb. I consider it a rare opportunity for anyone who wants to understand how AI works in production, far from the controlled environment of courses and notebooks.
CLICK HERE TO REGISTER FOR FREE!
Why the Feature Store Summit 2026 deserves your time
The opening session is delivered by Jim Dowling, CEO and co-founder of Hopsworks, associate professor at KTH Royal Institute of Technology in Stockholm, and author of Building Machine Learning Systems with a Feature Store, published by O’Reilly. I have followed his work for a long time, read the early-release version of the book, and continue to recommend the Hopsworks material to anyone who intends to take models to production.
If the term is new to you, a feature store is the infrastructure layer that computes, stores, and serves the input variables of a model, so that training and inference work with the same values. In this sixth edition, the discussion extends to AI agents, which depend on fresh, real-time context to produce reliable answers.
The talks I do not plan to miss
What caught my attention in the program is that a large share of the 17 talks covers real cases, presented by the people who built them. These are the four I do not plan to miss (all times in UTC):
- Uber (16:30): how to keep features consistent between training and production in large-scale search and ranking models.
- Red Hat (17:30): how to build agents on governed context by combining Feast, MLflow, and OGX.
- Spotify (18:40): how to look up a single key directly in the Parquet files of a data lake, accepting a little more latency in exchange for a considerable reduction in cost.
- Airbnb (20:30): how a generative recommender moved from daily batch processing to near-real-time updates.
The full agenda runs from 15:30 to 22:00 UTC (11:30 a.m. to 6:00 p.m. Eastern Time) and also includes teams from Adyen, Zomato, Zalando, and Delivery Hero. In addition, every registrant is entered into a draw for either a copy of the book signed by the author or €1,000 in Hopsworks platform credits.
Even if you cannot follow the entire event, I recommend registering and choosing the sessions closest to your own projects. Hearing how these teams reason about cost, latency, and consistency broadens your repertoire and helps you see the path between a trained model and a production system. Registration is free and available on the official Feature Store Summit website.
