Notes produced by
real engineering work.
Architecture decisions, integration seams, troubleshooting, operational lessons and design changes. Notes are evidence of how a system was understood, not just content to publish.
KubeMetal featured as an AI agent reference in XERJ Corpus Hub
A KubeMetal implementation separating Colima VM profiles is now available as an AI agent reference in XERJ Corpus Hub.
Read noteRead by evidence, not by card.
Why we consolidated beluga-manager into the beluga monorepo
Two repositories joined by pins had drifted 118 commits apart, and cross-repo changes could not land atomically. This post summarizes the decision to fold beluga-manager into beluga's manager/ directory, the alternatives, and the consequences, based on ADR-0005.
Soongsil Guest Lecture Recap: What Humans Should Hold in the Age of AI Agents
A write-up of 'AI Agents and the Role of the Human Developer', presented in Soongsil University's 2026 Software Special Topics 2. Silent failures drawn from my own incident logs, and a rule for what to hand off versus what to keep.
The Manual Steps I Kept Repeating on Every Cluster Rebuild, So I Built ClusterDeck
Why re-creating a Kubernetes cluster kept breaking SSH, kubeconfig, hosts, and CA trust by hand, and how a Profile fixed it
A CI Job Can Print FAIL and Still Report Success
How OpenForge's false-green detector grew from a swallowed `|| true` pattern to a second, structurally different shape, and the real narwhal/beluga bugs it caught along the way
Getting Started with eGovFrame Contribution: Lessons from Real PRs and Reviews
A 2026 field guide and slide deck on contributing to the Korean e-Government Standard Framework (eGovFrame), from spotting small issues to review handling and verification.
Proving Completion in Agent-Assisted Open Source Engineering
An implementation-backed record of treating resolved invocation, CI, runtime evidence, post-state verification, and portfolio publication—not code generation or merge alone—as proof of OSS completion
📚 Re-documenting 10 OSS Projects — From Source State to First Verified Success
A source- and evidence-based documentation review of Narwhal, Narwhal Portal, NFS Quota Agent, ldapium, kube-ready-box, ClusterDeck, Beluga, Beluga Manager, KubeMetal, and OpenForge, redesigned around Time to First Verified Success.
August 2026 OSS Implementation Update — What Is Actually on main
A main-branch, evidence-based update on Narwhal, NFS Quota Agent, KubeMetal, ldapium and OpenForge as of August 2026.
Why Does Code Survive While Engineering Decisions Disappear?
Why I introduced ADRs while managing shared OSS engineering standards in OpenForge, and how ADR → Standard → Automation → Adoption keeps decisions traceable.
How I Reworked the CNE UI with AI and a Design System
How Projects, Project Detail, Notes, Tech Digest, Docs, Speaking, and shared navigation were brought under one design language, with AI used as an execution partner.
GitHub Pages Deployment Concurrency After the Bun Migration
A practical record of how an older Pages deployment was cancelled after a newer main revision arrived, and how to interpret that state correctly.
Migrating from npm to Bun — From the First Failure to CI Evidence
A practical record of migrating dasomel.github.io from npm to Bun 1.4, including the first failed deployment, measured CI performance, and the removal of duplicate production builds.
Bun 1.4 Migration — Visual CI Evidence
A visual companion to the Bun migration case study, using measured GitHub Actions timings and pipeline diagrams.
AI-Era Open Source: Build the Bathhouse, Not the County Office
A review of my OSS projects through a user-friction lens inspired by the bathhouse metaphor: Narwhal, Beluga, KubeMetal, eGovFrame Launcher, nfs-quota-agent, and ldapium.
🧭 Operating Seven OSS Projects as One Engineering Portfolio — Building a Common Standard with OpenForge
How Narwhal, Beluga, KubeMetal, kube-ready-box, ldapium, nfs-quota-agent, and OpenForge became one OSS Engineering Portfolio with shared engineering and supply-chain standards.
🤖 How I Build Open Source with AI — Lessons from Developing Narwhal
How I use ChatGPT, Claude, Gemini, and Copilot together in real open-source development, and how AI is changing my role as an engineer while building Narwhal and other Cloud Native projects.
💬 Safe Repository Development with ChatGPT — Adopting Luna Chat Coder
How I applied Luna Chat Coder to a real technical blog repository to make repository work safer, reproducible, and easier to continue across ChatGPT sessions.
Redesigned the Site with Narrative Authority Design
The story of transforming a flat blog into a professional portfolio combining narrative and authority
Migrated My Blog to Astro
The story of revamping my blog from Jekyll to Astro
Kubernetes · Platform Engineering · Open Source · AI-assisted Development · Reliability · Architecture · Troubleshooting · Design System