The Future of Reproducible Research - Powered by Kubeflow

Explore the future of reproducible research powered by Kubeflow in this 35-minute conference talk by Trevor Grant at CNCF's KubeCon + CloudNativeCon North America 2022. Delve into the importance of reproducibility in scientific research and how Kubeflow, an open-source, cloud-native data science platform, revolutionizes the process. Learn about a peer-reviewed article published with a Kubeflow Pipeline, enabling easy verification and iterative improvement of results. Discover the benefits of publishing full pipelines alongside code and data for the academic community. Gain insights into the replication crisis, its causes, and consequences, as well as practical examples of reproducing research using Apache Spark, Kubernetes, and Kubeflow. Explore topics such as rapid testing for CT scans, de-noising images, and visualizing results using open-source solutions. Understand the structure of Kubeflow pipelines and their role in enhancing reproducibility. Conclude with actionable insights on implementing replication-friendly practices in research and leveraging Kubeflow for more efficient and transparent scientific workflows.

Syllabus

About the Narrator (me)
About Electric Elk
My Main Hustle This Month
REPRODUCIBILITY
The Replication Crisis: What Is It?
The Replication Crisis: Causes
The Replication Crisis: Consequences
My Experience Trying to Reproduce Research
Tower of Babel: Making Apache Spark, K8s, and Kubeflow Play Nice
Rapid Testing Needed - Desperately
CT Scans for Rapid Testing
Low Dose CT Scans
"De-Noising" Images
Data Source
Why (F)OSS Based Solution
S3 Buckets
Visualizing Results
Positive Outcomes: Cool Example for Book
Calls To Action: Assume Replication
Calls To Action: Assume they won't be using your laptop.
What is Kubeflow and Why Will it Help? Talk about Kubaflow pipelines a series of docker containers that execute stops then hand off data to next step
Setting up Kubeflow
Conclusion: Buy our book