Teenager’s Summer Project at Caltech Rewrites the Rules of Space Discovery

In a remarkable feat that highlights the convergence of artificial intelligence and astronomy, high school senior Matteo Paz has successfully unearthed 1.5 million previously undetected celestial objects using a computational model he developed during a summer research program at Caltech. The core of this breakthrough lies in Paz's ability to apply advanced machine learning techniques to vast datasets collected by NASA’s NEOWISE telescope, which has primarily been tasked with tracking near-Earth asteroids. Paz's journey began at Caltech's Planet Finder Academy in the summer of 2022, where his passion for astronomy ignited during hands-on experiences. With mentorship from scientists like Davy Kirkpatrick, Paz employed a Fourier and wavelet-based model to analyze nearly 200 terabytes of data. This ambitious endeavor was not only a personal challenge but a testament to the increasing role of AI in tackling complex scientific problems. The NEOWISE dataset, with its massive scale of 200 billion data points, presented considerable obstacles for traditional analysis methods. Paz's innovative AI model, aptly named VARnet, was specifically designed to sift through this data efficiently, detecting subtle variations in infrared signals that could indicate novel astronomical phenomena. This represents not merely a technical achievement but a critical expansion of what we understand about the universe. His publication of the findings in The Astronomical Journal marks a significant milestone, as it demonstrates the potential contributions of young scientists in peer-reviewed research. Paz's work has received national attention, underscored by his victory at the 2025 Regeneron Science Talent Search, where he secured a $250,000 prize. Beyond his immediate discoveries, Paz envisions broader applications for his AI model, including in fields as diverse as stock market analysis and environmental monitoring. His work not only contributes to our understanding of astronomical phenomena but also sets a precedent for how the scientific community can leverage the talents of young innovators. As we anticipate the full release of the VarWISE catalog in 2025, it is clear that Paz's contributions could have far-reaching implications for future astronomical research, opening doors to a wealth of new data and insights about the cosmos.