Kasra Zamanian

The night before the entrance exam for NODET, Iran's national program for gifted students, I was up until 3am, and not to study: there was a party at my place. I got in anyway, on understanding rather than knowledge: the questions were closer to an IQ test than a syllabus, and reading a problem has always come easier to me than memorizing the answer. I spent middle and high school there in mathematics and physics, then went on to Industrial Engineering at Tehran Polytechnic, where, between running the music association and the odd life-or-death moment on a hike, I still finished in the top 5% of my class. A merit scholarship took me to Bocconi in Milan for an MSc in Data Science, where I ran for course representative because students' voices were not getting through and someone had to say the uncomfortable things, graduated with a perfect GPA, and wrote a thesis that later became a €300k AI product.

The last five years I have spent building applied AI in production: computer vision, NLP, RAG, and agents. Along the way I have led teams when it was called for, founded a startup, felt the gravity of corporate hierarchy, and tasted the full ownership and autonomy of a mid-sized company.

The part that holds me now is the architecture layer: how AI systems are grounded, evaluated, and made reliable enough to act on, not just demoed once and forgotten.

Among all, I have never let work and books crowd out life itself. Never took my fingers off my piano and Chopin pieces for the last 18 years, played chess when my brain is fried (around 1800 ELO), a Brazilian Jiu-Jitsu blue belt, peaks and valleys from Sicilian volcanoes and the Dolomites to the Kungsleden above the Arctic Circle, and, most importantly, spinning a lot of people around the floor at salsa and bachata.

If you're building something hard in AI, I'd like to hear about it.