Blog
New essays, technical writing, and anything written directly for this version of the site.
Open blogProjects
Selected engineering work and representative public artifacts.
View projectsArt
Curated gallery pages built from a much larger historical archive.
See artLegacy Wiki
Recovered notes from multiple generations of the personal wiki, kept as a searchable archive.
Browse legacy notesRecent Posts
All postsWhen AI Agents Validate Each Other's Hallucinations: A Cautionary Tale from Production Research
A production research postmortem on how multi-agent analysis produced a cascading hallucination chain, why it was hard to detect, and what verification guardrails still require human judgment.
Markov: A Lightweight, YAML-First Workflow Engine for Chaining AI Agent Jobs on Kubernetes
An introduction to Markov, a YAML-first workflow engine that combines rule-engine gating, recursive workflows, and lightweight Kubernetes-native execution for AI agent job orchestration.
The Company That Invented Containers Five Years Too Early: The Rise and Fall of rPath
A history of rPath, Conary, and the company that anticipated containers, GitOps, reproducible builds, and immutable infrastructure years before the ecosystem was ready.
Your SDLC Has 36 Activity Types. Your Agent Can Handle 14 of Them.
A cognitive-load framework for deciding where AI agents help across the SDLC, where they need oversight, and where process redesign matters more than automation.
Beyond Binary Thinking: A Graduated Fidelity Spectrum for Kubernetes Security Testing
A decision framework for selecting the right Kubernetes security testing tier based on fidelity, cost, and developer feedback speed.