I build systems that build systems.

Open to platform · infrastructure · backend · AI-engineering roles — remote, EU.

crag — Run one command, get your CI, hooks, and AI coding configs synced automatically. Reads any codebase, writes a 20-line governance.md, compiles it to GitHub Actions, husky, pre-commit, Cursor, Copilot, Gemini, and 6 more. Zero dependencies. Deterministic — SHA-verified on every release. Stress-tested on 141 OSS repos. This is the public piece of the infrastructure described below.

Leyoda — AI-powered investor-startup matching platform. Spring Boot + Next.js + Python signal engine. Blue-green deploys, OAuth, production-grade CI/CD.

MetricHost — Distributed game server hosting platform. 11 Spring Boot microservices, Next.js frontend, self-managed multi-region Kubernetes with a flannel→Cilium (eBPF) migration, Terraform + Ansible IaC, Kafka event bus, Stripe billing, hibernation engine.

All three run on a custom AI development infrastructure I built in 4 days. I write code, run tests, and deploy to production from my phone — no laptop needed. The system discovers any project automatically, enforces quality gates, self-corrects its own knowledge, and optimizes its own cost. crag is the open-source foundation — the rest isn't public yet.

I don't just write code. I build execution infrastructure that makes everything I do from that point forward faster, cheaper, and operable from anywhere.

• Phone-first development — I operate full production systems from an iPhone through infrastructure I designed for mobile-first AI-assisted development
• Self-healing automation — deterministic hooks enforce safety at zero cost, memory self-corrects against source files, context survives compaction
• AI agent infrastructure — not using AI tools in default mode. Building the layer underneath them. crag is the first piece I've open-sourced

The result: park bench, phone, full project context loaded in under a minute, AI writes and tests code, I approve from a push notification, ships to production.

Systems over effort. Iteration speed over raw output. Ship real things, not demos.

Every project I touch gets production-grade infrastructure from day one — quality gates, security review, automated deployment, persistent memory. Not because I'm cautious, but because the infrastructure costs 20 minutes to set up and saves weeks over the project lifetime.

Something bigger is coming.