Summary
Mario Czacki is a Lead AI Architect and Product Manager with nine years of experience building foundation models, multi-agent systems, and the production infrastructure that turns research into deployed products. He combines a double degree in Computer Engineering and Data Science from UNAM with an MS in Data Science from NYU (Fulbright Scholar) and a track record of shipping compact, structure-first models that outperform larger scale baselines. At Kuona he scaled demand forecasting across 500+ servers and architected an agent platform from scratch, now leading teams building an energy-based foundation model for market panel data and a multi-agent analytical platform used by global retailers and manufacturers. He won first place in Yann LeCun’s Deep Learning course competition by using physics-informed, ~20K-parameter models to beat ~3M-parameter ViTs, underscoring his thesis that inductive biases often trump scale. Mario emphasizes determinism, auditability, and formal planning in agent design and publishes his thinking and experiments at mgczacki.github.io.
10 years of coding experience
4 years of employment as a software developer
Master of Science - MS, Data Science, Master of Science - MS, Data Science at New York University
Universidad Nacional Autónoma de México (UNAM)
High School Diploma, High School Diploma at Modern American School
English, Spanish