Justin
Garrison

Justin Garrison currently runs product at Sidero Labs, the company behind Talos Linux - an immutable, API-driven operating system built specifically for Kubernetes. That job title is a recent thing. The career trajectory behind it is not. This is someone who has been doing hard infrastructure work, writing about it honestly, and calling out nonsense publicly for over a decade.

The Disney+ story is the one that stops people cold. When Disney decided to launch its streaming service in 2019, they needed infrastructure that could go from zero to planetary scale - fast. Justin was on the team that built it. He managed 50,000-plus containers across four regions, navigated a pre-launch crisis when AWS called to say they had no available servers (yes, really), and helped the platform scale from its first subscriber to 50 million. That's not theory. That's a war story from someone who lived it.

Before that, he was accumulating the kind of knowledge base that only comes from doing real systems work over years. Math and physics undergraduate degree from Azusa Pacific, an MS in information technology, and then a string of infrastructure roles that built up to Disney's streaming ambitions. After Disney+ launched successfully, AWS came calling - and he spent three and a half years as a Senior Developer Advocate working on EKS, the Elastic Kubernetes Service. He was part of the team that launched EKS Anywhere, which let enterprises run the AWS Kubernetes experience on their own hardware. He also worked with the Karpenter project, the open-source node provisioner that can dynamically right-size your Kubernetes clusters. His documented results: one implementation dropped compute costs from $244,000 to $71,000 per month. That math is not subtle.

Platform engineering is where Justin gets particularly pointed. While the rest of the industry was sticking "platform engineering" labels on every internal tooling initiative, Justin published what might be his most shared blog post: a measured but firm argument that most "platform engineering" is just Central IT with a new name, a Kubernetes cluster bolted on, and the same organizational dysfunctions underneath. He's not wrong. That take lands because it comes from someone who has actually built platforms at scale - not someone selling a consulting engagement about platforms.

The move to Sidero Labs in 2024 makes sense in this light. Talos Linux is built around a premise Justin clearly believes in: that the operating system running your Kubernetes nodes should be immutable, minimal, and API-driven. No SSH. No package manager. No footguns. The system either works as designed or it doesn't - and if it doesn't, you replace it, you don't patch it. That's a philosophical position, not just a product feature.

Beyond the day job, Justin has built a media operation that would embarrass people who call themselves "content creators." The Ship It! podcast covers everything that happens after code is committed - DevOps, SRE, infrastructure, the messy middle of getting software to run in production. He co-hosts Fork Around and Find Out (FAFO - yes, the acronym is deliberate) with Autumn Nash, targeting the SRE and platform engineering community with a mix of expert interviews and frank takes on the state of the industry. His YouTube channel has cleared 80,000 subscribers. He was a regular contributor to The New Stack and How-To Geek. He ran a newsletter called 123dev for exactly 100 issues - one GIF, two comments, three links per issue, designed to be read in under ten minutes - then stopped it cleanly and wrote up what he learned.

The book is real and important: Cloud Native Infrastructure, co-authored with Kris Nova and published by O'Reilly. It was one of the first serious attempts to write down the patterns and practices for running infrastructure in a cloud native environment - not just the tooling, but the organizational and operational thinking behind it. Companies including Google, Amazon, and Netflix provided input for the case studies.

He is also an Oscar-winning contributor to Disney Animation films - a fact that appears on his CV with no further elaboration, which is exactly how you handle something like that.

Justin is an original chair of the Kubernetes SIG on-prem special interest group - the working group inside the Kubernetes project responsible for ensuring that Kubernetes works well outside of public clouds. This is substantive technical work. Special interest groups in CNCF projects are run by practitioners who do the work, not by committee appointees. Being a founding chair means he helped define what on-premises Kubernetes should look like before the industry had settled on an answer.

On a personal level: he reads. A lot. He finished 19 books in 2025 and published his reading list with notes. He runs his own Bluesky Personal Data Server from home on a Raspberry Pi - of course he does - and wrote the guide on how to do it. He is active on Bluesky and Mastodon as a deliberate choice to invest in decentralized platforms rather than cede all attention to algorithmically-driven ones. He has thirteen-plus professional certifications, including both Certified Kubernetes Administrator and Certified Kubernetes Application Developer. His GitHub username is @rothgar, not his name, which he has never felt the need to explain.

What ties it together is a consistent posture: do hard things, write about them honestly, skip the buzzwords. At a time when the cloud native ecosystem generates more noise than signal, Justin Garrison is one of the more reliable sources of signal.