Explore the challenges and solutions for cloud-native machine learning systems in this 26-minute conference talk from KubeCon + CloudNativeCon Europe 2021. Delve into the unique aspects of ML systems, including reproducibility, feature engineering, and explainability. Learn how to navigate pitfalls in interactive development, production monitoring, and DevOps practices for maturing cloud-native ML systems. Gain insights from industry experiences to address issues like GDPR compliance, high-velocity data science, and delivery challenges. Discover practical strategies for building robust, scalable ML infrastructures on Kubernetes, preparing you for day-two operations and beyond.
Cloud Native Machine Learning Systems at Day Two and Beyond
CNCF [Cloud Native Computing Foundation] via YouTube
Overview
Syllabus
Introduction
Agenda
Machine Learning Systems
Hidden Technical Debt
Reproducibility
Feature Engineering
Library Stack
Explainability
GDPR
High Velocity Data Science
Delivery Challenges
Inputs Outputs
The Big Picture
Taught by
CNCF [Cloud Native Computing Foundation]
