Bootable AI Containers with Podman Desktop

What if you could solve dependency hell by packaging your entire OS, AI model, and app into a single, bootable container image?

#1about 3 minutes

Introducing bootable containers and the CNCF donation

An overview of the session's agenda and the announcement of Red Hat's container tools, including bootable containers, being donated to the CNCF.

#2about 4 minutes

Solving OS management challenges with bootable containers

Bootable containers apply the immutable, standardized principles of application containers to entire operating systems to prevent configuration drift.

#3about 3 minutes

The workflow for deploying and updating bootable systems

A container file is used to build a bootable image, which is then converted to a disk image for deployment and can be updated by pulling a new version from a registry.

#4about 4 minutes

Building a bootable OS image with a container file

The `bootc` tool manages updates and rollbacks, while a container file defines the OS layers, packages, and applications, and you can even build a custom base image.

#5about 3 minutes

Why bootable containers are ideal for AI and ML stacks

Packaging AI/ML stacks as bootable containers solves dependency issues and allows models to run closer to the hardware for faster inference.

#6about 2 minutes

Simplifying the build process with Podman Desktop

Podman Desktop and its extensions provide a user-friendly interface to build a bootable container image and export it to various disk formats.

#7about 5 minutes

Demo: Developing a local AI code assistant application

A demonstration of building and running a Python-based AI code generation application as a standard container using the Podman Desktop AI Lab extension.

#8about 7 minutes