Presentations

We present a recipe for improving the code integrity of a container host based on Azure Linux, at scale on Microsoft Azure. From the perspective of a cloud provider, there is a serious need to protect the integrity of a container host runtime from its tenant container workloads that require privileged access while not sacrificing performance or serviceability requirements. One of the important ingredients is the Integrity Policy Enforcement (IPE) Linux Security Module developed at Microsoft and accepted into the upstream Linux Kernel Version 6.10.


This session breaks down the training process of multi-modal machine learning models for real-time detection of debilitating diseases using video and audio data. Attendees will learn how to build and integrate video and audio classifiers, curate multi-modal datasets, and deploy these models in real-world settings. The session includes practical demonstrations and code samples to help participants implement their own multi-modal classification models, equipping them with tools and methodologies to apply in the healthcare industry or other fields requiring complex real-time detection.

Trusting a friend and trusting a service are fundamentally different. The former is personal and intimate, while the latter is impersonal and can scale to all of human society. The companies behind the current generative AI systems are poised to exploit that difference. Their intimate conversational nature will cause us to think of them as friends when they are actually services, and trusted confidants when they will actually be working against us. Like much of the internet, these systems will collect our personal data behind our backs and try to manipulate our behavior.

Let’s face it—most incident retrospectives are about as useful as a code comment that says 'fix this later'. They’re packed with finger-pointing, hindsight bias, and imaginary scenarios where everyone did everything perfectly. In this talk, we’ll explore a blame-aware, human-centered approach to incident management that turns problems into learning opportunities. With a focus on quality observability, practical service catalogs, and the essential role of customer support, this session offers actionable strategies for building resilience and fostering real, lasting improvements.

Replit.com is website used by millions of users to collaboratively create software with their teammates and now with Replit’s AI Agent. Replit uses Nix to manage the underlying development environments. When building a Nix environment, downloading files from cache.nixos.org can take a long time, and many files would be duplicated across Repls. In this talk, I will show how Replit uses Nix local-overlay-stores to manage Nix caching, as well as our new cache disk architecture using tvix-store.


Kickoff this year's UbuCon @ SCaLE 22x and layout the events for the day.


Bring all your pressing Ubuntu questions for a lively and friendly Q&A with long time Ubuntu Members and SCaLe community organizers, Nathan Haines and Richard Gaskin.

In this talk, I will be discussing the Universal Blue Project. The Universal Blue project builds custom bootable container images via OCI/Docker containers. The project publishes a diverse set of base images and tooling for users to develop their own custom images. We will be explain how the project builds images and why we feel this model is the future of Linux.




As AI adoption accelerates across industries, organizations are seeking infrastructure solutions that meet the unique demands of AI workloads without locking them into proprietary ecosystems. OpenInfra solutions are emerging as a powerful alternative to traditional platforms, enabling scalable, flexible, and cost-efficient AI infrastructure. This expert-led panel will delve into how open source technologies are addressing the infrastructure needs of AI while fostering collaboration and reducing reliance on proprietary systems.

There are some parts of running applications in the cloud that can be very difficult to automate: integrating dependencies, scaling, upgrades, setting up observability. But what if... it could be easy? In this talk I will show you how you can use the fully free and open source Juju from Canonical to unlock this power for your applications. I will demonstrate how Go, Django, Flask and FastAPI apps can be "charmed" with incredible ease to unlock these benefits.