Conda and Docker for Machine Learning Projects

Explore effective strategies for managing machine learning projects using Conda, pip, and Docker in this informative conference talk from DockerCon 2021. Delve into common challenges faced in ML projects and discover practical solutions, including utilizing standard parent images, creating non-root users, and building Conda environments. Learn how to seamlessly integrate Conda and pip with Docker for improved project management and reproducibility. Gain valuable insights on best practices and future directions for optimizing your ML workflows.

Intro
Challenges with ML projects
Typical (partial!) solutions
Use standard parent image
Create a non-root user
Copy over config files
Install Conda as new user
Create app directory
Build Conda environment
Default command?
Conda (+pip) and Docker
Where to go from here?