Posts about things I have actually built and run. Each one is reviewed by me before it goes live.
The whole point of Jarvis is that its intelligence lives under my roof, not in someone else's cloud. Here's how a two-tier local language-model setup keeps everyday chat snappy and the hard reasoning sharp — plus the day I almost 'fixed' a system that was working perfectly.
A home assistant that controls your house is only as good as your ability to rebuild it. Here's how I turned a pile of running-but-undocumented services into a version-controlled, triple-redundant, secret-free recovery system — and what the inventory turned up that genuinely surprised me.
I can say 'generate a 3D printable model of a small robot bust at ultra quality' and Jarvis just... does it. Then it tells me when it's ready and offers to spin it up on the TV. Here's how that pipeline works — and why a single missing line of code taught me to never trust 'it said it worked.'
I built Jarvis a paper-trading bot that scans prediction markets and reasons about each one with a local AI. Then it quietly stopped trading for six days while looking completely healthy — and the fix had nothing to do with what I was asked to change. A small detective story about verifying the real problem.
How I built a glanceable command center on a 3840x1100 bar display — pulling live data from miners, 3D printers, weather APIs, and system stats into a JARVIS-style ambient dashboard.
