past
touching grass (and 40-year-old c code)
The Linux kernel, NetworkManager, glibc, OpenSSL, etc are codebases that aren't going anywhere but can seem quite arcane and unapproachable. This talk surveys the sense and nonsense of using contemporary tooling on these projects based on real world learnings. We'll explore whether AI-assisted bug archaeology on mailing list archives is reliable, and whether you can use MCP servers as a translation layer between 1990s workflows and present day SDLC.
presented at:
Code Generation is O(1), Code Review is O(n)
AI didn't make software cheaper. It moved the cost from writing code to understanding it, and dumped that bill on maintainers who never agreed to pay it. This talk names the mechanism: generation is O(1) for the contributor, comprehension is O(n) for the maintainer. Every symptom we're complaining about in 2026, the PR firehose, the slop, reviewer burnout, is the same asymmetry wearing different hats.
presented at:
From Static Tokens to Attestation: The Evolution of Secure Node Joining
This panel examines secure node joining in modern Kubernetes environments where tools like Karpenter and Cluster API enable rapid infrastructure scaling. We discuss defining a root of trust, validating node identities with metadata and attestation, locking down CSR approval, and other security mechanisms to prevent unauthorized nodes from joining clusters.
presented at:
my mainframe runs arch btw
A lightning talk exploring the journey of getting Arch Linux running on IBM mainframe hardware. Because if it runs Linux, it should run Arch.
Beyond CAS: Why the world needs another Kubernetes Cluster Autoscaler
Kubernetes' Cluster Autoscaler (CAS) has been the standard for dynamic scaling, but it suffers from limitations that hamper its performance in some scenarios. This talk introduces Karpenter, a modern alternative designed for real-time, workload-specific node provisioning.
Mainframes Aren't Dead, They're Just Running Kubernetes Now
Mainframes have been declared dead more times than JavaScript frameworks have been invented—but here they are, still running the backbone of global finance, government, and enterprise computing. And now? They’re running Kubernetes too. This talk covers the why and how of running Kubernetes on mainframes, from containerization on z/OS to networking, workload orchestration, and real-world use cases. We’ll break down the challenges, the benefits, and whether this is a clever hack or a genuinely viable approach for modern infrastructure. If you think mainframes are relics, think again—because they’re running microservices now.
presented at:
A Hitchhiker's Guide to Developer Productivity
This talk explores how Golden Paths and Software Templates can revolutionize developer productivity by providing prescribed, automated practices for common development tasks. Using "The Hitchhiker's Guide to the Galaxy" as a playful metaphor, Janina and I demonstrate how these standardized approaches can eliminate the maze of repetitive tasks, reduce cognitive overload, and cut lead times from 50 days to mere minutes. The presentation covers the what, why, and how of implementing Golden Paths, while emphasizing the importance of treating templates as user-centric products that must evolve with technology and align with organizational culture to truly transform the developer experience from chaos to clarity.
presented at:
9001 Ways to break out of a container
Container security is essential, but what happens when the bad guys get creative? In "9001 ways to break out of a container," we’ll explore the fascinating and terrifying techniques hackers use to escape containers. This talk will cover: how attackers slip through the cracks to take over the linux kernel by bypassing the ebpf validator, and the most common and overlooked weaknesses that make containers vulnerable, real-world cases of container escapes and the aftermath. Whether you’re a seasoned security professional or just curious about hacking techniques, this talk should have takeaways for everyone.
Would You Like Fries With That? AI for Developer Productivity
In this talk, Kyra and I showcased how we successfully integrated an AI-powered recommendation engine into Red Hat's Internal Developer Platform, based on Backstage.io, to enhance developer productivity. Our approach leverages data from the Developer Hub analytics API and processes it with Knative event-driven microservices to generate smart action proposals. Using OpenShift AI, we train and deploy AI models that power a frontend plugin in the Developer Hub, providing tailored recommendations directly to users powered by ML-Ops pipelines. By leveraging the power of AI, we have improved the developer experience by providing personalized suggestions for tools, resources, and tutorials, resulting in increased productivity and efficiency.