I build and ship production AI systems — not slide decks. Where most AI projects stall as polished demos nobody uses, I deliver systems that run in production and stay maintained. I work directly with the people who do the work, so the system that ships is the one already running in their hands — minus the parts a machine does better. What I build: Agentic process automation • automation on n8n with AI agents, plus custom microservices (APIs, scrapers, document converters) where low-code ends. - Document intelligence & RAG • turning hundreds of thousands of PDFs, scans and mixed formats into a searchable knowledge base for both people and AI agents. - Agentic software development • full applications and backends built agentically (Claude Code + MCP), always under the control of architecture, specifications, tests and a multi-agent CI. - MCP, fine-tuning & architecture • MCP servers, model selection and fine-tuning, with security and cost control built in. I'm the creator and maintainer of n8n-mcp • the open-source tool that lets AI agents build n8n automations on their own. It's used by 100,000 engineers, including teams at Deutsche Telekom, PayPal, MIT and NYU, with 21,800 stars on GitHub. Selected work: World Bank Group, the Saudi Ministry of Industry, EnergyHouse, Thedy & Partners, CFOpro. Previously Head of AI at Boldare (2020–2024) and creator of the "Praktycznie o AI" YouTube channel. If you want to move from AI experiments to architecturally coherent, maintainable production systems, let's talk.