The Three-Line Cron Job That Should Have Existed Years Ago
I finally set up automated offsite backups for this blog after fourteen years of winging it, and the whole thing took about twenty minutes once I stopped overthinking it.
Tutorials
Tutorials
What Yesterday's AWS Outage Made Me Go Check On My Own Site
Yesterday's big AWS outage took out half the apps on my phone, so I spent the afternoon checking whether my own tiny site had the same dumb single points of failure.
What the Windows 10 ESU Enrollment Actually Asks For
Notes from actually enrolling an old Windows 10 machine in Microsoft's Extended Security Updates program the day after support ended, including the part about the "free" option nobody explains well.
A Weekend Fixing Twelve Years of Bad Image Habits With Claude Code
A weekend spent finally using Claude Code to clean up twelve years of unoptimized blog images, and the one lesson about scoping tasks narrowly that I keep having to relearn.
Feeding My Entire Blog Archive to GPT-4.1-nano
OpenAI quietly shipped GPT-4.1 and two smaller siblings straight to the API this week, so I spent an evening dumping fourteen years of this blog into the cheapest one just to see what a million-token context window is actually good for.
Vibe Coding From the Terminal, Three Weeks In
Three weeks into using terminal-based AI coding tools like Claude Code and Cursor, some real notes on what works, what's overhyped, and the one habit that's saved me from a bad diff more than once.
So Now It's Called "Vibe Coding"
Karpathy coined "vibe coding" a couple weeks ago and it's everywhere now. Here's how I actually use Cursor and Claude day to day, and the one bug that taught me to keep reading my diffs.
Running DeepSeek R1 Locally Instead of Downloading the App
Skipping the DeepSeek app and the stock market panic entirely, I just ran R1's distilled models locally with Ollama on a three-year-old MacBook. Here's what that actually looked like.
I Ran DeepSeek R1 on My Own Laptop and My Fan Has Not Forgiven Me
A hands-on walkthrough of pulling DeepSeek R1's distilled models onto a regular laptop with Ollama, plus honest notes on speed, heat, and whether it's actually any good.
Getting DeepSeek R1 to Think Out Loud on My Laptop
A weekend spent getting DeepSeek R1's distilled models running locally with Ollama on a regular MacBook, and why watching it think out loud in a terminal beat the Stargate headlines.
The ThinkPad Gets a Second Life as a Chatbot
I spent part of the holiday break getting a local chatbot running on a six-year-old ThinkPad, and it actually works well enough that I've mostly stopped opening ChatGPT.
Fixing My Bluesky Feed So It Stops Sucking
A practical walkthrough of the custom feeds, starter packs, and mute settings that finally made my Bluesky timeline worth opening every morning.