Blog
1655+ articles — Page 54 of 69
What Breaks First When AI Moves from PoC to
The first thing that breaks moving AI from PoC to production on Kubernetes is the assumption that a successful demo equals a production-ready platform.
2 min read
AI Platform First 90 Days on Kubernetes: A Pragmatic Roadmap
Build the first production-ready AI path, then expand. A pragmatic 90-day roadmap for teams starting AI infrastructure on Kubernetes.
4 min read
Autoscale AI Inference on Kubernetes: HPA and KEDA
Inference autoscaling is harder because every extra replica means another GPU and cold-start penalty. Custom metrics like tokens/second are essential.
4 min read
The Hidden Cost Drivers in AI Workloads on Kubernetes
The obvious cost is GPUs. The hidden costs make GPUs sit idle. Three buckets of AI waste — compute, data-path, and operational — and fixes.
4 min read
The Architecture Decisions That Are Hardest
You can replace tools later. Unwinding the wrong tenancy model or GPU scheduling policy is much harder. Get these decisions right early.
4 min read
KubeCon 2026 Takeaway: AI on Kubernetes Has
KubeCon EU 2026: AI on Kubernetes moved from experimentation to industrialization. Inference is the new center, production scale is the benchmark.
3 min read
Building a Multi-Tenant GPU Platform: It's an
Multi-tenant GPU platforms fail when treated as purely technical. The harder problem is agreeing who gets what. Here is the operating model.
3 min read
AI Observability on Kubernetes: The Three
AI observability must connect infrastructure signals, model behavior, and business outcomes. Three layers and how to instrument them.
4 min read
Platform + SRE + ML Teams: The Production
Platform, SRE, and ML teams need a clear contract — shared accountability with explicit boundaries. Define ownership by outcome, not by tool.
4 min read
The Biggest Mistakes Teams Make Scaling AI/ML
Treating AI workloads like bigger microservices is the fastest way to fail. GPU sharing, experiment sprawl, and premature optimization mistakes.
3 min read
Kubernetes Recipes Book Giveaway at KubeCon
Kubernetes Recipes book giveaway at KubeCon EU 2026 was a massive hit. Signed copies at the vCluster booth with co-author Grzegorz Stencel.
3 min read
5-Spot Scheduler: Cut Bare Metal Kubernetes Costs by 70%
FINOS 5-Spot Machine Scheduler is a Kubernetes operator for time-based scheduling of physical infrastructure. Born at RBC for bare-metal cost optimization.
5 min read
On Stage at KubeCon EU 2026: Multi-Tenant GPUs on Bare Metal
Photo essay from KubeCon EU 2026 — from title slide to SR-IOV code, the full story of multi-tenant GPUs on the biggest Cloud Native stage.
4 min read
KubeCon Europe 2026: Where 13,350 People Became a Community
KubeCon Amsterdam 2026: 13,350 people from every corner of the world building something bigger than any single company or project.
6 min read
KubeCon Amsterdam 2026 in Numbers: Cloud Native and AI
19.9M cloud native developers, 255B inference market by 2030, 13,350 attendees. KubeCon EU 2026 numbers tell the AI convergence story.
4 min read
Packed Room at KubeCon Europe 2026:
780 attendees packed the room for my KubeCon EU 2026 talk on multi-tenant GPUs on bare metal OpenShift AI — recap and key takeaways inside.
4 min read
Speaking at Conferences: From Ansible Berlin
From Ansible Berlin 2023 to FOSDEM 2025 and KubeCon EU 2026 — a look back at my conference speaking journey across the Cloud Native ecosystem.
5 min read
The Developer Burden: 2026 Reality Check on
75% of developers lose 6-15 hours weekly to tool sprawl. Platform engineering is the only way out of the developer burden crisis.
7 min read
Jensen Huang Says Engineers Should Spend $250K on AI Tokens
NVIDIA CEO Jensen Huang claims engineers should consume 250K in AI tokens. Why he is partly right, mostly self-serving, and what it means.
5 min read
Cloud Native Rejekts 2026 Amsterdam: MC Experience
Co-hosting Cloud Native Rejekts 2026 with Julia Hahn — 300+ engineers, deep technical talks, and the best B-side conference before KubeCon.
4 min read
Meet Artem Lajko: GitOps Pioneer Launching
Artem Lajko, Head of Platform Engineering at iits-consulting, launches Kubara — an open-source Kubernetes platform built on GitOps.
4 min read
Meet David Parry: AI-Powered Code Reviews and the Future of
David Parry, Principal Architect at Qodo, brings AI-driven code reviews to open source. Catch his KubeAutoDay talk on Agent MCPs.
3 min read
Meet Pepijn Oomen: Building Europe's
Pepijn Oomen is a Kubestronaut building Nordic Cloud at Redpill Linpro — sovereign, GDPR-compliant, 100% green energy Kubernetes platform.
5 min read
KubeCon Europe 2026 Companies: The
Meet the companies making waves at KubeCon EU 2026 — sovereign cloud, AI code reviews, and open-source Kubernetes frameworks.
2 min read