The SAME Project - A Cloud Native Approach to Reproducible Machine Learning

Explore a cloud native approach to reproducible machine learning in this conference talk from KubeCon + CloudNativeCon North America 2021. Dive into the Self-Assembling Machine Learning Environment (SAME) project, a Kubernetes and Kubeflow initiative aimed at addressing challenges in reproducing state-of-the-art machine learning models. Learn about the project's goals, initial public release, and future directions. Discover how SAME tackles issues of quick ramp-up, seamless collaboration, and efficient scaling in machine learning environments. Gain insights into the project's use of Kubernetes for composability, portability, and scalability. Understand the concept of the SAME File and its role in declaratively executing runs. Examine the user experience for data scientists and ML engineers, and explore the broader community needs addressed by this innovative approach to reproducible machine learning.

Introduction
Vocabulary
The Problem
Two Options
Kubernetes
composability
portability
scalability
containers and Kubernetes
The SAME Project
Goal
How it might work
Running in production
Setting the context
Declaratively executing runs
Recap
User Experience
The Same File
Data Scientists and ML Engineers
Need in the Community