Watch the Recordings: Mirantis Talks at KubeCon EU 2025

Last month’s KubeCon EU 2025 in London was by far Mirantis’ most successful KubeCon in multiple respects. We not only presented our largest number of talks to date at a KubeCon conference (6 talks, including our first keynote), but we also garnered extensive press coverage — with 34 articles across more than 30 publications — earning us the greatest share of voice among all vendors at the conference.  

Our talks at KubeCon EU this year showcased the depth and impact of our work across the cloud native ecosystem. From platform resilience and multi-cluster management to driving inclusivity in open source, Mirantis team members delivered powerful sessions that resonated with developers, platform engineers, and thought leaders alike. In case you didn’t get a chance to attend KubeCon, or you missed some of our sessions, here’s a roundup of the key ideas, innovations, and insights our speakers brought to the stage.

Redefining MLOps at Scale on Top of Kubernetes

Mirantis VP of Open Source Strategy and Technology Randy Bias’s keynote highlights how open source innovation has challenged previously dominant players like OpenAI, with projects like DeepSeek demonstrating that open source can be the way of the future. Similarly, Kubernetes has the opportunity to succeed in managing the surge of AI-powered applications, but faces a major challenge in infrastructure sprawl. 

This is where open source k0rdent comes into play: a super control plane designed to deploy, run, and manage Kubernetes clusters at scale across private clouds, public clouds, VMs, bare metal, and edge environments. Running atop the lightweight k0s distribution and integrated with OpenStack and CNCF ecosystem components, k0rdent enables rapid deployment of AI inference engines anywhere in the world within minutes.

Key Takeaways:

• Open-source innovation is disrupting established leaders and can similarly take the lead in cloud infrastructure.

• As AI applications continue to rise in popularity, Kubernetes must tackle sprawl and scalability issues to remain viable.

• k0rdent is a super control plane that can manage and deploy Kubernetes clusters efficiently across varied environments. The demo illustrates deploying an AI model within minutes, showcasing k0rdent's speed and versatility.

Read Randy’s latest blog, A New Era of Open Source Governance.

Abstracting Multi-Cluster Topologies With CAPI and Linkerd for Internal Developer Platform

Mirantis Consulting Architect William Rizzo delves into the idea of resiliency as a foundational requirement for cloud native applications, particularly in internal developer platforms (IDPs) where uptime and failure tolerance are non-negotiable. Traditional approaches such as stretched clusters and highly available components reduce failure risks but still centralize the point of failure. 

A more robust solution involves abstracting resiliency into the platform itself using tools like Linkerd and Cluster API (CAPI). By leveraging multi-cluster and federated service architectures, resiliency can be delivered as a core platform capability that is not exposed to developers. This model enables applications to remain operational even when individual clusters fail. Components like k0rdent’s k0rdent Cluster Manager (KCM), k0rdent State Manager (KSM), and k0rdent Observability & FinOps (KOF) help manage the lifecycle, state, and observability of clusters at scale, forming the backbone of a resilient platform.

Key Takeaways:

• Resiliency should be a built-in platform capability, not an afterthought.

• Traditional resilience methods (e.g. stretched clusters or high-availability setups) are helpful but still expose single points of failure like control planes.

• Federated service architectures using Linkerd allow applications to operate across multiple clusters and failure domains seamlessly, improving fault tolerance.

• Cluster API (CAPI) facilitates scalable, automated provisioning and management of clusters across varied infrastructure environments.

• k0rdent’s architecture (KCM, KSM, KOF) abstracts cluster lifecycle, service state management, and observability into modular, reusable components for multi-cluster resiliency.

• Developers should not be exposed to cluster topologies, but platforms should still communicate deployment environments transparently to support optimization.

Solving Real-World Edge Challenges With K0s, NATS, and Raspberry Pi Clusters 

Mirantis Senior DevOps Engineer Prashant Ramhit outlines a real-world marine conservation project where edge computing technologies were used to monitor coral reef health offshore. To overcome challenges like vendor lock-in, high costs, and limited customization, Arkatech NGO built a resilient, low-cost system using Raspberry Pi devices housed in waterproof enclosures and powered by solar panels. They deployed a software stack based on k0s, an open source, single-binary Kubernetes distribution, combined with NATS for real-time messaging and a Golang-based monitoring application. The solution enables autonomous data collection of critical scientific metrics like temperature, pH, salinity, and nutrient levels, supporting coral restoration efforts. The project demonstrates the viability of running Kubernetes clusters on minimal, ruggedized hardware in harsh, remote environments, and promoted broader adoption of open source tools for sustainable, scalable edge computing.

Key Takeaways:

• Low-cost, resilient edge computing is achievable using open source technologies like k0s, NATS, and Raspberry Pi, even in harsh, remote marine environments.

• k0s provides a lightweight, production-ready Kubernetes platform that simplifies deployment and management without external dependencies—ideal for constrained edge scenarios.

• Real-time environmental monitoring for coral restoration was successfully enabled through autonomous, solar-powered buoys collecting critical scientific data offshore.

• Open source, customizable architectures empower organizations to innovate beyond traditional vendor limitations, supporting scalable solutions across environmental and industrial use cases.

Building Thriving Communities in Platform Engineering: Collaboration, Innovation, and Growth

William Rizzo joins a panel with Bart Farrell (Learnk8s), Courtney Nickerson (Nirmata), Kelly Revenaugh (Testkube), and Matteo Bianchi (GitHub) to explore how thriving platform engineering initiatives depend not just on technical excellence but on intentional and inclusive communities. A great platform isn’t useful unless it’s used, and communities that foster trust and shared ownership are a great stepping stone to adoption. Building community means starting with people: interviewing teams, listening actively, and avoiding assumptions. Engaging internal users early and often — through one-on-one conversations, lightweight feedback loops, and collaborative roadmap planning — ensures that the platform addresses real needs and edge cases that might otherwise be overlooked. Diverse roles and backgrounds strengthen community outcomes; successful platform teams often include not just software engineers but also professionals from QA, UX, developer relations, and customer-facing roles who can advocate for usability and accessibility. 

Community building also means meeting users where they are, with practices like holding open office hours, running participatory hackathons, and embedding platform engineers within product teams to increase visibility and feedback flow. Thorough documentation is also valuable as both a tool and a community artifact, especially when maintained as a shared responsibility. Ultimately, communities thrive when people feel heard and platforms succeed when they grow out of collective trust.

Key Takeaways:

• Adoption depends on community engagement. Even the best platforms fail without active involvement from the people they’re built for, which makes feedback essential.

• There is no one-size-fits-all approach to build vs. buy. Decisions vary based on team size, legacy systems, and compliance needs; hybrid strategies often work best.

• Diverse roles contribute to successful platform engineering. Backgrounds in QA, UX, developer advocacy, and beyond bring valuable perspectives that improve adoption and usability.

• Strong feedback loops require ownership. Feedback should be actively collected, tracked, and turned into action. This process also builds trust for future feedback loops. 

• Community is about presence and trust. Personal interaction (e.g. interviews, office hours, and cross-functional collaboration) helps build trust and encourage participation.

• Documentation and communication matter deeply. Well-maintained, collaborative documentation and stakeholder-focused metrics help platforms scale sustainably.

See You at the Next Event!

KubeCon EU 2025 showcased not only the depth of Mirantis’ technical innovation and expertise, but also our commitment to community, collaboration, and platform engineering best practices. If you didn’t get a chance to meet us in London, we’ll also be at KCD Istanbul on May 23-24 and DevOpsCon Berlin on June 17-18. Hope to see you at one of the events!