I am a Principal Research Data Scientist at iSpot, where I work on Agentic AI, Large Language Models, and multimodal ad creative analysis.
We led the full journey of iSpot SAGE AI for creative analysis—from scrappy experiments and proof-of-concepts to a polished, production-ready system handed off to engineering. Along the way, we pushed the frontier of agentic LLMs, streamlined generative AI workflows for advertising analytics, and built novel ad-creative intelligence architectures powered by proprietary data and cognitive models.
I received my Ph.D. in Mathematics from UIC in 2021, advised by Prof. Alex Furman. My research focused on dynamical systems, ergodic theory, and hyperbolic geometry. My work has spanned both pure mathematics and computational applications, from horoball packing problems to neural network modeling.
You can find my publications here, or explore the Packing Problems and Complex Dynamics sections.
Professional inquiries: robert [dot] kozma [at] ispot [dot] tv
Academic correspondence: robert [at] kozma [dot] ai
GitHub: github.com/rkthijs