Building Dynamic Machine Learning Pipelines with KubeDirector

Explore the intricacies of building dynamic machine learning pipelines using KubeDirector in this 55-minute webinar from the Cloud Native Computing Foundation (CNCF). Dive into a practical example of an ML pipeline designed to predict travel times based on taxi ride data. Learn how to develop a complete ML pipeline using Kubernetes and KubeDirector, covering the processes of training, registering, and querying your model. Discover the power of KubeDirector's "Connections" feature in maintaining an up-to-date ML model. Follow along as experts Tom Phelan, Kartik Mathur, and Donald Wake from Hewlett Packard Enterprise guide you through topics such as stateful applications, the KubeDirector operator, Jupyter Notebook integration, and pipeline configuration. Gain insights into advanced concepts like service mesh and network layers, and understand the differences between stateful and stateless applications in the context of ML pipelines.

Introduction
Overview
Application types
Stateful application
KubeDirector operator
Basic machine learning pipeline
KubeDirector Application
Green Blocks
Example Pipeline
KubeDirector Cluster
Connection
Training App
ML Training
Jupyter Notebook
Connections
What is the Jupiter Notebook
What is the Jupiter Lab
Building the pipeline
Config map
Cons
Deployment
Connection feature
Running the pipeline
Whats next
Contact us
Stateful vs stateless
Questions
Why dont you handle that one
What is ECM
What is Spark Operator
Questions and comments
Service Mesh
Network Layer
Wrapup
How do I get one of those shirts