Thirteen years from Photoshop to production servers.

I started in 2013, doing graphic design for friends who needed flyers, brochures, and logos. I was the one with Photoshop and a Canon 550D — I'd photograph someone's business and have a printed flyer for them the next day. Over the next few years I moved from print to web — first hand-coded HTML and CSS, then customizing themes for DLE, Drupal, WordPress, Joomla. I drew about 50 site designs along the way.

By 2018 I was building landing pages on no-code site builders — Wix, Weblium, similar platforms. By 2019 I was offering full packages: logo, Figma prototype, deployed site, external server for documentation and assets. Around the same time I started running my own VPS — partly because a shared host refused to update PHP for me, partly because I'd just earned $700 in tokens running a node for the Evmos testnet and realized I could host my own things instead of renting fragments of someone else's machine. That's when I started learning Nginx, SSH, server hardening — the boring stack that quietly runs everything.

In 2022, when russia's full-scale invasion of Ukraine began, I lost every client I had. Some businesses closed, some stopped replying. Design wasn't on anyone's mind. I spent most of a year doing nothing — not even thinking, really. Just a kind of shock about whether anything I'd been doing still mattered. The right word for that state is depression. There isn't a softer one.

In 2023 I met someone online — turned out we had mutual contacts — and started drawing site designs for his agency, around 15 of them. He kept telling me to drop Apache and WordPress and learn Laravel. After four months of him repeating that, I gave in. GPT-4 at the time could barely write working code, so learning Laravel meant actually understanding it — which, in hindsight, was the best version of that learning curve to be on.

In 2024 I went deep into trading. I learned every classical indicator, every system, every theory — and concluded most of them were noise built for markets that no longer exist. So I started writing my own, first in Pine Script for TradingView, then in C# for cTrader when Pine couldn't do what I needed. Around 30 failed attempts in, the indicators started working.

By the end of 2025, AI could finally write real code, and I installed Claude Code on my VPS. That's where I am now: a team of seven in one person — designer, developer, sysadmin, project manager, copywriter, marketer, QA. Working end-to-end on whatever I need to build.

What I care about

Shipping things that work, not things that demo well.

AI as a typing assistant, not an engineer. The judgment about what to build, why it breaks, and how to fix it stays human.

Owning the whole stack. From the Figma file to the systemd service — same person, same accountability.

Drawing the whole interface before AI writes the code. It's the difference between getting what you asked for on the first try and spending three months debugging what AI thought you meant.

Currently

Building zborys.dev as a home for everything I've made and learned.

Shipping SENU toward a stable release — open source, free, looking for sysadmins who'd actually use it and tell me what's broken.

Open to projects worth the time.

Trading taught me that any system you can't ship — and that doesn't produce clear visible results — is just theory.