About
Mikulas Tomanka
I lead cloud reliability for a multi-tenant platform: on-call, incidents, and the team that runs them. I was one of the platform's founding engineers too, so the system I keep alive is one I helped build.
I built the regional deployment units the platform ships into new regions, all defined as code, and then stayed on to run them in production. The day-to-day is the unglamorous side of software: capacity, cost, failure modes, and the boundaries between teams. A lot of it is making sure an incident only happens once. What I've learned is that reliability comes down to ownership more than technology. Systems fail at the seams between teams, and those seams usually come from the org chart, not the architecture.
Off the clock I build small software on my own. One is a tax calculator for self-employed people stuck with a hostile set of rules; another is a financial-independence planner that runs entirely in the browser. I don't pretend they're startups. They show that one person with real domain knowledge and a tight scope can ship production-grade tools, and these days I build them with an AI pair that turns a spec into code faster than I can second-guess it. The harder and more interesting part is the guardrails those agents run inside.
That has changed how I see both halves of my work. When writing code stops being the slow part, the job becomes deciding what to build and catching the machine when it's confidently wrong. So far that judgment is still on me.