Feature Store
Summit 2026Register ▸

[ Past editions / 2024 ]

FSS/24

18 sessions · 25 speakers · Feature Store Summit 2024

[ WATCH / PLAYLIST ]

[ SPEAKERS // 25 ]

[ AGENDA // 18 ]

Oct 15, 06:30 AM UTC · 10 min · Kickoff
› Beyond Feature Stores: Data for AI in Real-time, Batch, and LLMs
We will explore the journey from traditional Feature Stores to a more comprehensive framework that integrates these diverse data types.
Oct 15, 06:40 AM UTC · 20 min · Presentation
› From Feature Store to AI Lakehouse
Hopsworks introduces the AI Lakehouse - an extension to the Lakehouse with MLOps capabilities, real-time data support, LLMs and more.
Oct 15, 07:00 AM UTC · 20 min · Presentation
› Uber's GenAI Oncall Co-Pilot Journey
Uber discusses how they built a GenAI Bot to boost on-call efficiency and minimize downtime.
Oct 15, 07:20 AM UTC · 20 min · Presentation
› Chronon, Airbnb's Open Source Feature Engineering Framework
Airbnb talks about Chronon - an open-source solution for ML practitioners, ensuring online-offline consistency.
Oct 15, 07:40 AM UTC · 20 min · Presentation
› Serving Real-Time Features at Etsy
Learn more about Etsy’s Rivulet, a real-time feature store, enhancing ML performance by ensuring feature freshness.
Oct 15, 08:10 AM UTC · 20 min · Presentation
› Shepherd: High-Scale, Low-Latency Machine Learning with Flink at Stripe
Stripe explores Shepherd's architecture: Flink, tiled data storage, and an automated control plane for faster feature development.
Oct 15, 08:30 AM UTC · 20 min · Presentation
› From Dearth to Discovery: Anomaly Detection using Generative AI
Supriya discusses leveraging generative AI models for anomaly detection, crucial for early fraud detection and risk management.
Oct 15, 08:30 AM UTC · 20 min · Presentation
› Real-time Feature Serving for Online Inference
Learn how Chalk built its just-in-time online feature store to enable realistic ML use-cases and serve requests in under 5 milliseconds.
Oct 15, 08:50 AM UTC · 20 min · Presentation
› The Snowflake Schema Data Model comes to Feature Stores
Learn how Hopsworks now supports the Snowflake Schema Data Model, enabling more features via foreign keys in online tables.
Oct 15, 09:10 AM UTC · 20 min · Presentation
› Feature Store as a Service: How Intuit's Feature Store Service Boosts Developer Productivity on One Intuit Platform
Discover Intuit's strategy for a millisecond-ready Feature Store, enabling developers to build faster, data-driven apps.
Oct 15, 09:40 AM UTC · 15 min · Presentation
› Immutable KV Store on Cassandra
Learn how Uber's Michelangelo team uses Cassandra for online prediction, its limitations, and how an immutable store offers a solution.
Oct 15, 09:55 AM UTC · 15 min · Presentation
› TurboML’s platform to leverage fresh data for ML
Learn how TurboML's platform overcomes the challenges posed by real-time data that enable fresher features, faster models and more.
Oct 15, 10:10 AM UTC · 15 min · Presentation
› Enabling Low Latency Fraud Detection with Real-Time Feature Engineering
Learn how to build real-time fraud detection pipelines with Quix Streams, a Python library, for faster, simpler feature computation.
Oct 15, 10:25 AM UTC · 15 min · Presentation
› Real-Time Fine-Tuning of Embedding Models for Improved Retrieval with Hopsworks and Bytewax
Learn to optimize continuous embedding models with Hopsworks and Bytewax for real-time improvements in content recommendation and more.
Oct 15, 10:50 AM UTC · 15 min · Presentation
› Large-Scale Embedding Feature Generation at Uber
Explore how Uber uses embeddings for ML systems, covering their lifecycle, from creation to deployment, and their impact on performance.
Oct 15, 11:05 AM UTC · 15 min · Presentation
› How Roche’s Data Platform accelerates Feature Engineering through Generative AI
Discover how Roche’s data platform uses GenAI to transform feature engineering, enhancing healthcare analytics.
Oct 15, 11:20 AM UTC · 15 min · Presentation
› Fennel's Primitives for maintaining Data & Feature Quality
Learn seven key primitives from Fennel that ensure data and features are reliable and trustworthy for high-quality ML models.
Oct 15, 11:50 AM UTC · 10 min · Wrap up
› Wrap-up
Jim Dowling wraps up the event.

◂ FSS/23 [ archive ]FSS/25 ▸