This blog is the thinking, not the changelog

· Roman Miakotin

  • introduction
  • resilience
  • engineering
  • energy
A small back-facing figure seated at a writing desk, looking down at an open notebook with a small hand-drawn diagram; a pencil rests across the notebook and a cobalt blue ceramic mug with rising steam sits to its right.

Some software gets to assume the lights stay on. The kind I work on doesn’t.

It started with home energy systems in places where the grid is not a given — where power is scheduled, interrupted, or simply gone, and the people relying on it are still trying to work, keep food cold, and stay online through it. But the grid was never really the subject. The subject was building systems meant to hold where they’re not allowed to fail: decisions made under uncertainty, software that has to stay smart while everything around it degrades, the difference between reacting and deciding. That same problem shows up one level higher, too — in the way software itself gets built, now that the tools writing it can’t be fully trusted either.

This blog is where I write that thinking down.

A few things it is not. It isn’t a changelog — I’m building something in this space, with different priorities than the obvious tools, but here the product is context, not the subject. It isn’t a tutorial mill or a hardware-review feed. And it isn’t thought-leadership theatre: no “5 trends,” no manufactured contrarianism, no post that should have been a tweet.

What it is: engineering essays, one layer up. Not “how I built X” but how to think about the class of problem X belongs to — resilient systems under real-world stress, energy and edge and IoT, the architecture of decisions, and how software itself gets built now that AI is in the loop. Opinionated, specific, and willing to show the wrong turn, because the wrong turn is usually the part worth reading. If I make a claim, I’ll argue it. If I reject something popular, I’ll say why on the engineering, not the vibes.

Who it’s for: the person who builds things and wants the reasoning, not the press release. If you’ve ever watched a system run flawlessly in the demo and fail at 2 a.m., we’ll get along.

I won’t promise a posting cadence I can’t hold. I’ll promise this instead: every post earns the time it asks for. Remove the product entirely and there should still be something here worth reading — that’s the bar I’m holding myself to.

Start anywhere. Then tell me where I’m wrong.