About
Building software with AI agents — not as a thought experiment, but as a daily practice across startups, enterprise, and everything in between.
Patryk Szymański
I'm a software engineer who's spent the last few years figuring out how to build real software with AI agents — not demos, not toy projects, but production systems that serve customers, train teams, and ship on deadlines.
I wear a few hats. I'm CTO and co-founder at Skillmint, where we're building Minty — an AI-powered platform for recruitment workflows. I built the entire backend using my own AI development framework called Loom, which I've been iterating on for over a year and shipped to v1.0. At STX Next, I work as a senior backend developer on OpenStack infrastructure, run AI bootcamps that teach developers how to work with coding agents, and lead a team of 18 developers.
This blog is where I share what I'm learning along the way. Not recycled takes about "AI will change everything" — but the specific, hard-won lessons from doing it every day. What works. What doesn't. And why the gap between AI demos and production systems is where the real engineering happens.
What I Actually Do
Building Minty (Skillmint)
AI-powered recruitment platform. I lead the technical side — backend architecture, infrastructure, and the AI workflows that power the product. Currently in pre-launch, about to go live with customers.
Loom — AI Development Framework
An open-source-adjacent framework for building software with AI agents. Skills-as-programs, deterministic orchestration, context management. It's what I use to build Minty, and it's the source of most of my opinions about how AI-assisted development should work.
AI Bootcamps & Team Leadership (STX Next)
I design and run hands-on bootcamps that teach developers how to work with coding agents — Cursor, Copilot, Claude Code. I also lead a team of 18 developers, which gives me a front-row seat to how AI tools actually change (or don't change) how teams build software.
Infrastructure & Homelab
I run a 4-node Proxmox cluster at home — the kind of infrastructure that teaches you more about reliability, networking, and recovery than any cloud certification. Occasionally I write about it.
What I Write About
AI-Native Engineering. Agent workflows, skills-as-programs, the economics of AI-assisted development, code comprehension when machines write the code, and the tools and patterns I'm building into Loom.
Building Minty. The startup journey — architecture decisions, shipping under pressure, the reality of going from prototype to production with AI as your development partner. Build-in-public without the hype.
Systems & Infrastructure. Occasional deep dives into homelab architecture, self-hosting, and the infrastructure decisions that keep things running when you don't have a cloud budget.
Let's Connect
Want to talk about AI-assisted development, building with agents, or swap infrastructure war stories?