Kubernetes for Humans

In this episode, Itiel Shwartz kicks off a series on MLOps, LLM, and GenAI in Kubernetes. Starting with Alexei Ledenev, who has over two decades in software development and deep experience in cloud architecture and distributed systems. He shares his journey from CoreOS Fleet to his current role on the Platform Team at Doit. The conversation focuses on tackling the complexity of Kubernetes, which Alexei notes can be overwhelming even for experienced DevOps engineers. He discusses how he developed the idea to leverage AI assistants and the Model Context Protocol (MCP) to access and execute tools like kubectl. This concept creates a "translator between AI and the Kubernetes environment", allowing users to troubleshoot complex cluster issues or quickly create ad hoc testing environments using natural language. They also explore the challenges of implementation, such as hallucination, and how providing context helps the AI self-correct. Looking ahead, Alexei predicts that infrastructure is moving towards self-aware and self-healing platforms that integrate AI deeply.

Meet Dustin Kirkland, VP of Engineering at Chainguard. Dustin shares his fascinating 26-year journey in the tech industry, from IBM and two stints at Canonical to roles at Google (working on GKE), Apex, and Goldman Sachs, eventually leading him back to engineering at Chainguard. Discover how Chainguard is helping secure the software supply chain, focusing on building secure containers primarily for Kubernetes. Learn about the critical problem of software vulnerabilities (CVEs) and how Chainguard's products, including their hardened container images with a zero CVE goal and accelerated patching SLAs, address this challenge. Dustin also introduces their newer initiatives, like Chainguard Libraries, which secures open source dependencies and builds hardened virtual machines (kernels) for Kubernetes worker nodes. He explains Chainguard's automation-driven approach, which allows them to monitor upstream fixes and rapidly rebuild and retest images and libraries.

Join Itiel as he chats with Gianluca Mardente, a Principal Engineer at Cisco Systems. Gianluca shares his path to tech and Kubernetes, including his work history and the inspiration behind his open-source project, Sveltos. They dive into the significant challenges of managing a large fleet of Kubernetes clusters – ensuring consistency, handling upgrades, and coordinating resources across different clusters. Learn how Gianluca's project tackles these 'Day-2 Operations', offering solutions that go beyond simple cluster creation and enable seamless integration of various open-source tools like Crossplane. Check out Sveltos: https://projectsveltos.github.io/sveltos/

In this episode, Sid, CEO of Hathora, discusses building game infrastructure, specifically for hosting dedicated servers. He shares how Hathora tackles the challenges of running stateful, low-latency, high-throughput workloads that reconcile player actions up to 60 times per second. Sid explains their approach using Kubernetes to manage compute across bare metal and cloud VMs, leveraging technologies like Talos and Civo's Omni. He also details the technical hurdles they face with Kubernetes networking (like hostport issues), achieving sub-3-second pod startup times, and optimizing container image pulls, offering insights into the unique demands of the gaming industry and Hathora's growth.Sid is the CEO and co-founder of Hathora, the premier tool for studios to globally scale dedicated servers for their multiplayer video games. Sid comes from a deeply technical background, having led infra teams responsible for $100m+ annual cloud budgets. His team at Hathora is working with some of the most anticipated games of '25 and beyond, including Frost Giant’s Stormgate, Omeda’s Predecessor, and 1047 Games’ Splitgate 2.Follow Sid on LinkedIn: https://www.linkedin.com/in/dsiddharth/ Learn more about Hathora at hathora.dev

In this special KubeCon edition episode of the Kubernetes for Humans podcast, Itiel meets with Lukas Gentele, CEO and co-founder of Loft Labs. Discover why Lukas believes multi-tenancy is a major obstacle in Kubernetes adoption and how Loft Labs is tackling this challenge through innovative projects like vCluster, a "super famous project" for running virtual Kubernetes clusters. Learn about the journey that led to vCluster's creation, stemming from the challenges of running a multi-tenant PaaS offering. Lukas also shares insights into the adoption of vCluster, including surprising stories of users running thousands of virtual clusters. The episode further explores Loft Labs' philosophy on balancing open source and commercial offerings. The highlight? The announcement of vNode, a brand new commercial product for virtualizing nodes and achieving node-level security to separate vClusters even further. Understand the limitations of existing solutions like micro-VMs and GVisor and how VNode offers a lightweight alternative using Linux kernel magic. Finally, Lukas shares his prediction on the future of virtualization, suggesting a shift away from traditional VMs towards lighter-weight solutions like VNode. Check out Loft Labs: https://www.loft.sh/Learn more about vNode: https://www.vnode.com/

In this episode, Itiel sits down with Kasper, a Developer Advocate at Dash0. Kasper discusses his background, including his extensive experience building platforms on Kubernetes at Lunar, a Nordic challenger bank. He shares insights into the challenges and successes of using cutting-edge cloud-native technologies in a regulated banking environment. Kasper also touches upon his involvement in the CNCF community as an ambassador and co-chair for KubeCon. The discussion explores Lunar's philosophy of leveraging open-source projects and their journey of adopting Kubernetes and other cloud-native tools. Kasper also explains his move to Dash0, emphasizing his passion for observability and the potential of Open Telemetry.

In this episode of Kubernetes for Humans, we speak with Ahmed Bebars, a Principal Engineer at the New York Times and a CNCF Ambassador, who offers a unique perspective on cloud native technologies. Ahmed recounts his professional journey in accounting before transitioning into the technology sector, leading to his current deep involvement with the Kubernetes ecosystem. He shares his initial introduction to Kubernetes almost a decade ago, recognizing its capabilities in container orchestration. Ahmed provided his insights on the appropriate use cases for Kubernetes, emphasizing the importance of identifying the problem before selecting a solution. The discussion also covers service mesh technologies such as Istio, along with Ahmed's perspective on their complexities and benefits. Looking ahead, he shares his thoughts on the evolution of the Kubernetes ecosystem, including the potential impact of GenAI. Follow Ahmed on Linkedin and Youtube Don't miss Ahmed's upcoming KubeCon talk!