Didur Labs LLC

The starting point is always the real problem, not a product to sell. That means asking what actually needs to happen, what constraints are non-negotiable, and what a working result looks like before choosing any tool or technology.

The bias is toward systems you own and understand: local-first where it makes sense, open platforms over closed ones, and clear seams between components so the whole thing stays serviceable.

Hard problems rarely respect the boundary between hardware and software. Didur Labs works across electronics, firmware, mechanical design, networking, software, and AI—so a solution can move fluidly between a PCB, a piece of firmware, a network design, and an automation without handing the problem off at every boundary.

That range is the point. It allows the work to address problems that do not fit neatly into a single trade or category.

A working prototype teaches more than a perfect plan. The approach favors building something real and testable early, letting the physical world correct the assumptions, and iterating toward a dependable result.

The measure of success is simple: does it solve the problem, and does it keep working after the engineer leaves?

A system nobody can understand is a liability, however clever it is. Every project aims to leave behind clear documentation—how it is built, how it is wired, how it is configured, and how to service it.

The goal is systems that can be supported, extended, and handed off, rather than black boxes that depend on their builder forever.

Ongoing interests include locally hosted AI and intelligent agents, computer vision for practical automation, low-power embedded devices, and deeper integration between custom hardware and open automation platforms.

Much of this work lives in The Lab—experiments and prototypes that may or may not become products, but consistently produce useful knowledge.