IDEAS, DECISIONS & LESSONS
Notes on AI, architecture and building useful products.
7 articles
08 Oct 2026 · 4 min read
devops
Kubelet's failCgroupV1 field defaulted to true in 1.35. A cgroup v1-only node won't start anymore unless you explicitly override it. Here's how to check today.
Read more10 Aug 2026 · 5 min read
aws
AWS EKS engineers published the real root cause of slow CUDA/PyTorch image pulls on Aug 10, 2026: it's serialization, not bandwidth. Two containerd fixes now shipping in EKS Auto Mode — chunked range-request downloads and concurrent layer unpacking.
Read more22 Jul 2026 · 7 min read
ai-agents
CNCF's July 2026 paper makes the case that Kubernetes is the natural convergence platform for AI agents. Here's why they're correct — and how to run AI agents on Kubernetes today with proper isolation, policy enforcement, and observability.
Read more30 Mar 2026 · 6 min read
ai
The CNCF just made Dapr Agents v1.0 generally available. After years of AI agent frameworks focused on intelligence, here's one that focuses on survival — durable workflows, state management, and production-grade reliability on Kubernetes.
Read more10 Mar 2026 · 24 min read
deployment
Part 8 of 8 — The complete deployment guide: Docker multi-stage builds, Kubernetes orchestration, CI/CD with GitHub Actions, zero-downtime deploys, go-live checklist, production monitoring with Prometheus/Grafana, and the operational runbook that keeps voice AI running at scale.
Read more28 Feb 2026 · 18 min read
architecture
Part 10 of 12 — From 10 concurrent interviews to 10,000. LiveKit SFU mesh, stateless agent workers, Kubernetes auto-scaling, regional deployment, and the infrastructure patterns that handle hiring season surges.
Read more27 Feb 2026 · 9 min read
devops
Docker and Kubernetes explained like you're having coffee with a friend. No jargon walls — just real analogies about shipping containers, apartments, and why 'works on my machine' is everyone's least favorite sentence.
Read more
