AI systems architect and independent researcher form Poland. I build multi-agent reasoning systems and regulation-aware AI stacks for high-stakes domains (medical, AML, financial crime). My parallel reasoning platform Yggdrasil tested on the GAIA benchmark.
My writing sits at the intersection of AI architecture, cognitive science, and the philosophy of mind. I am particularly interested in how humans and machines compress knowledge into usable form, where that compression introduces error, and what it would take to build better correction mechanisms into the interfaces we rely on.
I approach most questions empirically rather than dogmatically. I am skeptical of clean theoretical frameworks that do not survive contact with messy real-world systems, including in AI alignment.