The Origin Story of Eurisko, the Most Advanced Math/CS Track in the USA (www.justinmath.com)

Eurisko is the origin story of a super-advanced high-school math/CS track run inside Math Academy from 2020–23 that taught students to reproduce academic AI research and build core systems from first principles in Python. Rather than relying on libraries, pupils implemented neural networks and backpropagation, game trees, evolutionary algorithms and more—then used those tools to replicate and extend papers. The program claims to have been the most advanced U.S. high-school track of its kind, and its first cohort is now in college with early but notable outcomes: accelerated graduate degrees, publications, science-fair wins, and a standout example who published research that identified 1.5 million previously unknown space objects, boosted a NASA mission’s potential, and won the Regeneron Science Talent Search. For the AI/ML community the significance is twofold: pedagogically, Eurisko demonstrates the power of deep, hands-on algorithmic training (coding learning primitives from scratch) to produce researchers who can both reproduce literature and innovate; institutionally, it offers a pipeline for earlier, more capable contributors to the field. The team is now converting their manual talent-development curriculum into a scalable online system, aiming to broaden access and accelerate the emergence of skilled, research-ready contributors to AI and related domains.