If you want to talk, submit your presentation idea here.
The Python-Centric approach of Hopsworks 3.0
Hopsworks feature store bridges the gap between an enterprise data ecosystem (e.g. data warehouses, data lakes, streams) and its machine learning capabilities. At its core, Hopsworks provides data scientists access to reliable and fast feature data to develop, train and productionize models.
Feature Stores for Sub-ML
In this talk, Venkata Pingali shares Scribble Data's perspective the Sub-ML space, and how feature stores must change in order to keep up with the growing number of Sub-ML requirements in organizations.
Building an open-source learn-to-rank engine on top of a feature store
In this talk, Roman and Vsevolod talk about Metarank, an open-source personalization tool that can optimize the ordering of items to boost important business metrics like engagement, CTR, and conversion. Metarank uses an internal embedded feature store under the hood, and they discuss why they made a decision to reinvent the wheel and not stick to existing solutions.
Reliable Feature Engineering Pipelines with Hopsworks
With the rise of ML driven products, the health and correctness of MLOps and feature engineering pipelines has become crucial for a great user experience. In this presentation Fabio Buso discusses some of the strategies users can adopt on Hopsworks to test their feature engineering pipelines. From unit testing all the way to data validation and monitoring. He shows how users can integrate these strategies with the newly released GIT integration feature and orchestration capabilities that Hopsworks provides.
Building a SQL-Centric Feature Store
Jordan from Continual AI takes a look at the pros and cons of building a SQL-centric feature store. He explores how such a system can be leveraged in a powerful operational system that helps to automate ML workflows. He also compares and contrast such a system to other ML Platforms, whether more developer-oriented notebook tools, or the more recent onset of MLOps-oriented tools.
High Performance Feature Processing using Snowpark
One of the biggest challenges around rapidly operationalizing ML models is the complexity of automating feature pipelines in a secure and reliable way at scale for both model retraining and inference. Simon Field walks through how the Snowflake capabilities can be used for feature stores and he gives customer examples of how to use Snowflake alongside other technologies as part of an MLOps toolchain.
RonDB, a LATS database as a Feature Store in Hopsworks
RonDB is a LATS database (low Latency, high Availability, high Throughput, scalable Storage). This presentation describes the role of RonDB in the Hopsworks Feature Store and what features that makes it suitable as the optimal choice for a Feature Store.
Overcoming data infrastructure limitations with a new paradigm for machine learning
Operating machine learning products requires trading off between the priorities of a data scientists’ training environment with those of data engineers’ production environment. In this session, come learn how to bridge the batch/streaming gap through a series of innovations in data processing. You’ll gain an understanding of how to build a feature compute engine that lets you get from idea to production in weeks rather than months.
Feature Store for Multi-tenant and Multi-app in Salesforce
Alex Araujo and Weiping Peng from Salesforce discuss how they are building ML applications in a multi-tenant environment. As they scale up to the number of teams and the number of applications building on our ML platform, Feature Store became a critical requirement for these teams to build and collaborate.
Data Engineering for Event-based Models
Ben Chambers from Kaskada surveys existing technologies that enables time-travel and discuss how a feature engine with time travel allows you to easily build and operate models from event-based data.
Feature Stores at Spotify
Over 345 million Spotify users rely on Spotify’s great recommendations and personalized features in 170 different markets around the globe (with 85 of these markets launching in the first part of 2021. With the massive inflows of data and complexity of production use cases, defining a unified approach to ML is challenging.
In this talk, Aman and Martin give an overview of the challenges we face with building a central ML Platform at a highly autonomous organization, and our approach of adoption by incentive. They dive deeper into our history with feature tooling, the foundation they are building now and where they are headed with a Feature Marketplace strategy at Spotify.
Real-Time Predictions with a Feature Store
Fabio Buso from Logical Clocks discusses how to leverage the online feature store can to compute and ingest features and make them available to operational models making real-time predictions, with low latency and preventing skew between the training and serving features.
Keying Redis Towards a Cost Efficient Online Feature Store
Arbaz Khan talks how Doordash nearly tripled our cost reduction by employing Redis Hashes efficiently and caring for how we serialize different kinds of feature data.
Feature Stores for Scientists
Davor Bonaci and Max Boyd from Kaskada discuss the importance and challenges of using event-based and time-series data in feature engineering.
Feature Management at Intuit
Monte Zweben, CEO at Splice Machine, presents the the motivation and key differences in the company's ML architecture and new approach for a Feature Store, with a single engine.
FeatureForm: The Virtual Feature Store
Simba from Featureform describes how a virtual feature store can provide all the value of a traditional feature store without having to re-write a single feature or dealing with vendor lock-in.
Disclaimer: This meetup included a presentation from Airbnb's Zipline feature store. Due to internal policies the company has requested to be removed from the recording.
Feature Engineering at Uber with Michelangelo
Atindriyo Sanyal, Technical Lead at Uber, discusses how feature engineering is done at Uber with the first feature store, Michelangelo.
Simplifying the Feature Store
Monte Zweben, CEO at Splice Machine, presents the the motivation and key differences in the company's ML architecture and new approach for a Feature Store, with a single engine
Hopsworks Feature Store 2.0: A new paradigm - Bridging the Gap between Training and Production with Feature Stores
Jim Dowling from Logical Clocks and Davor Bonaci, from Kaskada dive deep into what feature store is in machine learning, what different problems it solves, what to expect in 2021, and more.