Hopsworks - The Python-Centric Enterprise Feature Store

This talk demonstrates how to use the Python-centric Hopsworks feature store to build feature pipelines, create training data, validate features, and serve real-time machine-learning predictions. It follows an end-to-end intrusion detection example from raw data through model serving.

Syllabus

Introduction
What is a feature store
How does a business add value
ML Operations
Feature Store for ML
How do I get data into it
How do we write feature pipelines
Writing data to the feature store
Feature views
Python
Feature Pipeline
Feature Quality
RealTime Features
Feature View
Point in Time Correct Join
Preengineered features
Incremental training
Saving training data
Hopsworks properties
Hopsworks demo
Demo project
Data preview
Feature group activity
Training data sets
Intrusion detection example
Sample log click logs
Models
Model Data
Connecting to the Feature Store
Creating an API Key
Accessing the Platform
Predicting