I'm a senior software engineer who builds AI systems for production. Twenty years of engineering experience, now applied to AI agent orchestration and retrieval-augmented generation.
I started writing code professionally at KULeuven in 2006, working as a software developer and system administrator. After that I moved to Vancouver to work as a software engineer at SignalQ, a Canadian telecom. I came back to Belgium in 2015 to work at a couple of startups (Les Etalages, Social Seeder), getting my hands on the messy parts of building software products. In 2016 I joined Wolters Kluwer as a senior engineer, then moved to Napoleon Sports & Casino in 2019, eventually leading the technical team. In 2025 I joined Cuez as an AI & Software Engineer.
I've kept Protoku running alongside my main roles since I founded it in 2018. It's where I do my own work, my experiments, and where my AI engineering practice has grown.
I started studying AI seriously in 2017 with Udacity's Artificial Intelligence Nanodegree, well before the current wave of large language models. Around the same time I organized the PHPLimburg meetup (2018–2020).
For the past year I've been deep in production AI engineering. Most recently I built the orchestrator agent for the Google + BBC AI Agents demo at IBC2025, which won the Broadcast Tech Innovation Award. The work made me realize how much practical RAG and agent engineering knowledge is locked in private codebases and not written down anywhere useful.
So I'm writing it down. Retrieval-Augmented Generation: An Engineer's Guide to Building RAG Systems with Your Own Data is currently in flight on Leanpub. I also write at herczeg.be/blog about the gap between AI demos and production systems, and I speak at meetups and conferences when the topic fits.
I travel a lot with my wife. We've lived for stretches in Vancouver, Los Angeles, Japan, New Zealand, and Austria, with shorter trips in between. When I'm not on a plane, I'm training for a marathon or playing competitive tennis.
