I build systems.
Then I keep watching them.
Architecture is only the starting point. The lab follows how systems evolve through commits, releases, regressions, incidents and the knowledge produced by operating them.
Standards → Baseline → Platform → Capabilities → Workloads
The map is intentionally read top-to-bottom on mobile: engineering rules create a reproducible baseline, the platform integrates shared capabilities, and workloads expose the seams that need verification.
Systems under observation
Open a system to inspect problem, architecture, engineering decisions, development pulse, proof and lessons.
All systemsA reproducible and verifiable Kubernetes Internal Developer Platform
A local Kubernetes data platform connecting Kafka, CDC, Flink, Iceberg, Trino, Superset, and Airflow, with its unified control plane (manager/) in the same repository
An Apple Silicon hybrid MLOps desktop app separating Kubernetes control from native macOS MLX compute
Knowledge with provenance
A note matters more when the system, decision or failure that produced it remains visible.
Open KnowledgeWhy I forked Harbor for multi-arch images after Bitnami's free images stopped, why I shut it down when v2.16.0 shipped official arm64, and what 11 months of running a fork taught me
A record of installing the pstack skill collection, removing it, and moving only its new ideas into the global prompts of three coding agents, with the selection criteria and how loading was verified
Notes on kube-ready-box v0.3.0: a rebuild that took shipped Vagrant boxes to zero fixable CVEs, the HCP Vagrant shutdown discovered during upload, the move to GitHub Release + Pages, and a weekly scan for vulnerabilities that are not patched yet.
A write-up of the talk at Kubernetes Korea Group Meetup, Session 2 on October 28, 2026. Project roles seen through a Day-0/1/2 frame, the OpenForge verification system, outside-developer cases, and failures of an AI agent team.
How the judgement evolved
Framework → DevOps → Cloud Native → Platform → OSS → AI-assisted engineering. The tools changed; the operating principles became clearer.