Getting a Local Model Running on My Ancient Desktop
I finally got a small local AI model running on my old desktop instead of assuming it was too weak, and it took less than two hours once I stopped procrastinating.
I finally got a small local AI model running on my old desktop instead of assuming it was too weak, and it took less than two hours once I stopped procrastinating.
I spent a Saturday clicking every outbound link in my oldest posts and found out how much of the early-2010s internet has just quietly vanished.
I spent a few hours pruning my bloated RSS feed list and it turned into an accidental meditation on which blogs last and which ones quietly die.
Fifteen years into this blog and my entire backup strategy is still a shoebox of dying hard drives under my desk. A dispatch from three hours I will not get back.
After years of half-hearted attempts, I finally self-hosted FreshRSS on my Synology, ditched Inoreader, and discovered a third of my old feeds were dead anyway.
I finally audited every subscription hitting my card and found $71 a month of stuff I forgot I was paying for, including a QR code generator app.
How I finally migrated 41,000 family photos off Google Photos onto a self-hosted Immich server, including the metadata mistake that cost me a full re-import.
A weekend spent rebuilding my side project's deploy setup around plain git push, why the bare-repo trick still works, and where it fell apart.
A Sunday-night cleanup of twenty-two hoarded domains turns into a small confession about buying domains instead of finishing projects.
After a near-miss deleting my only backup folder by hand, I finally set up real automated backups for this blog — restic, nightly cron, and a weekly restore test.
I got tired of paying for uptime monitoring SaaS, so I built a stupidly simple one myself with a cron job, curl, and one plain HTML file.
How I set up FreshRSS on a home server to replace my Twitter doomscrolling habit, with the docker-compose config and the one trick that actually made it stick.