Field Notes
Short, direct entries — updated whenever there's something real to say.
Latest
Why I Build Instead of Just Studying It
Most of what I know about hardware didn't come from a lecture — it came from frying a servo, figuring out why, and rewiring it correctly the second time. I could read about current draw and power smoothing for a semester and understand less than I learned the first time a LiPo pack sagged under load and I had to actually solve it.
That's the whole premise behind Loki and JARVIS: build the real thing, on a real budget, and let the gaps in my knowledge show up as actual failures instead of abstract weak spots. It's slower than following a tutorial. It's also the only way I've found that actually sticks.
Why Local, Not API
Almost everything people call "AI" right now is a phone call to someone else's server. That's fine for a lot of use cases — but it means the moment the connection drops, the company changes its pricing, or the product gets discontinued, whatever you built on top of it stops existing.
JARVIS is my answer to that: a core model I trained myself, running on hardware I own, with no dependency on an external API to function. It doesn't need to be the smartest model in the world. It needs to be mine — fully inspectable, fully offline-capable, and still working in five years regardless of what any company does with their product roadmap.