I spent almost 20 years collecting roles across the software delivery lifecycle on purpose. Product owner, delivery manager, release manager, development manager, QA manager. The idea was simple: if I understand every part of the chain, I can lead the whole thing. Since AI coding tools got serious, I've been all in. I use Claude Code, Cursor, and Copilot daily, and in the past few months I built and open-sourced five full-stack apps while working full-time.
I’ve sat in most of the chairs. Here’s the highlights. If you want the full CV, just ask.
I manage 10-17 engineers across two verticals for a Fortune 100 networking vendor. Test automation frameworks, switch and router OS development. Before that, I ran a 13,000-project enterprise migration at SII for the Hitachi Energy / ABB separation. Three years, 16 TB of data, zero critical downtime.
Ran a 45-person engineering org. Led an agile transformation that took on-time delivery from near zero to 80% in one year. Right-sized the team from 45 to 30 without slowing delivery. Migrated to cloud, cut infrastructure costs 60%, grew conversion 10% year over year.
Head of IT at LowCostTravelGroup where I built a 40-person org from scratch. QA manager and release manager at Sabre, running operations for 300+ engineers. Product manager at Open-E. Project manager at Alan Systems delivering bwin.com. I picked these roles on purpose because I wanted to understand the full picture.
Why this blog exists
Writing forces me to think clearly. If I can explain what I built and why, I actually understand it. If I can’t, I have more work to do. That’s the whole reason this blog exists.
Reading about AI tools is one thing. Shipping five apps with them while working full-time is another. I learn by doing, and I write about what I find.
20 years of running teams means I think about costs, trade-offs, and whether someone will actually use what I'm building. That habit doesn't turn off when I'm coding at midnight.
Most people post about finished products. I write about the 11pm architecture decisions, the provider that turned out to be too expensive, and the features I cut.
Agile transformation I led at Fru.pl. 45-person engineering org, one year. We also right-sized the team from 45 to 30 and cut the monthly run rate by 150K PLN without slowing down.
Hitachi Energy / ABB corporate separation. 16 TB of data, multiple time zones, three years of coordination. Zero critical path downtime.
From product roadmap decisions I owned as CPO at Fru.pl. I ran the backlog for a major Polish e-commerce platform and prioritised based on what the data actually said, not what felt right.
SwiftSpeak, AI Studio, LiveTranslator, Intonavio, DictAll. Built with AI coding agents, open-sourced, shipped on evenings and weekends.
Built with Claude Code, Cursor, and Copilot. Open-sourced. Shipped on evenings and weekends while working full-time as an engineering manager.
I wanted better dictation than what Apple ships. So I built an iOS keyboard and macOS menu bar app with 10+ transcription providers, AI formatting, and RAG-powered context memory. BYOK model, so users pay for their own API keys.
Started as a way to manage my own AI agent workflows. Grew into a full MCP control plane with JIRA-like project management, 50+ tools, real-time telemetry, and code quality metrics.
My parents speak Polish, my wife’s family speaks Arabic. I got tired of being the interpreter during family visits, so I built a real-time translation app in a week. QR code room sharing, multi-provider STT, sub-2s latency.
A singing practice app that turns YouTube lyrics videos into interactive training sessions. Stem separation, real-time pitch detection, accuracy scoring. Built because I wanted to practice guitar vocals and nothing good existed.
Lightweight macOS menu bar dictation app. Multi-provider speech-to-text at about $3.74/year instead of $144/year for commercial alternatives. Built it because I dictate a lot and didn’t want to keep paying for it.
Writing and speaking
Build logs, cost breakdowns, things that didn’t work. I write here because it forces me to think clearly about what I’m actually doing with AI tools.
Webinar on AI coding agents without losing engineering standards (CodiLime, 2026). Previously spoke at AgileByExample and Bitspiration about agile transformations.
"Transformation to Platform Engineering" on the CodiLime blog (2025). I write about what I know from doing the work, not from reading about it.
Life beyond laptops
I live between Poland and Egypt. The things I do outside of work aren't separate from how I lead. They just use different muscles.
Guitar and drums. I also picked up Middle Eastern duff percussion after moving to Cairo. It's the thing that clears my head after a long day of code reviews.
I completed a three-year programme in Traditional Chinese Medicine and acupuncture. Green belt in karate. I think about my body the same way I think about systems at work: figure out how it works, keep it running well.
I volunteered to design and run a year-long yoga school for my local community. 12 students, monthly weekend intensives, full curriculum I built from scratch. Leading outside of work keeps the skills sharp.
If you’re building with AI agents, leading engineering teams through this shift, or just want to compare notes, I’d like to hear from you.
Say hello