Engineering deep dives on infrastructure, distributed systems, and AI/ML — from homelab experiments to enterprise architecture.
A practical case study of implementing centralized AI agent context management in a real .NET plugin repository. Demonstrates the AGENTS.md pattern with minimal per-tool stubs for Kiro, Claude Code, GitHub Copilot, Gemini CLI, and OpenAI Codex.
Implementing the REST API and admin dashboard for the Jellyfin Media Integrity Scanner plugin. Covers the ASP.NET Core controller, scan result queries, manual scan triggers, and the HTML/JavaScript dashboard with library health overview. Part 4 of a 6-part development series.
Building the core scanning engine for the Jellyfin Media Integrity Scanner plugin. Covers .NET 8 plugin scaffolding, FFmpeg process wrapper, bounded concurrent task queue, cross-platform path resolution, and production-safe error handling. Part 3 of a 6-part development series.
The tanaka boot-mirror migration (2x HDD to 2x 128GB SSD), run one day after Part 4's edgar migration with every lesson from that article applied up front: SSH trust, Ceph repos, ceph.conf/monmap cleanup, the SAN bridge, and the USB SMART quirks file. Covers a real hardware troubleshooting detour before the OS even mattered, a new pvecm add finding, and where the cluster's remaining spinning-rust…
A real execution of the Part 3 planned-migration procedure against a live node with both ZFS boot mirror drives degraded. Covers the SMART trigger, a confirmed UEFI dead end on this hardware, gaps in the original checklist (SAN network, Ceph mon/mgr recreation, USB SMART quirks), a hostid mistake that broke the next boot, a live OSD failure caused by the USB gap, and full verification.
Architecture and design decisions for the Jellyfin Media Integrity Scanner plugin. Covers plugin vs. script tradeoffs, two-phase scanning strategy, SQLite schema design, I/O throttling model, and CephFS/NFS storage considerations. Part 2 of a 6-part development series.
Introducing the Jellyfin Media Integrity Scanner plugin project — a production-safe tool for detecting corrupt, truncated, and damaged media files in your Jellyfin library without impacting playback performance. Part 1 of a 6-part development series.
How to validate a machine-readable resume with JSON-LD structured data against Google Rich Results Test, Schema.org validators, ATS resume parsers (Jobscan, Resume Worded), and AI agents. Includes testing methodology and results.
A pattern for using AI coding agents to generate self-contained HTML reports and dashboards, then publishing them to a Jekyll site on GitHub Pages for stable URLs, git-tracked history, and zero-infrastructure sharing. Covers the markdown-first content workflow, report design standards, GitHub Actions automation, and when to use HTML vs markdown posts.
Diagnosing and fixing the Jekyll Run VS Code extension crash in multi-root workspaces on macOS. A debugging story covering the misleading null toString TypeError, blind alleys through VS Code settings and plugin source code, the real root cause (macOS GUI PATH inheritance), and the complete solution using rbenv with Ruby 3.3.