
High-School AI Maps the Infrared Sky, Flagging 1.5 Million Hidden Cosmic Signals
Seventeen-year-old Matteo Paz, mentored by Caltech researchers, built VARnet to scan NASA’s NEOWISE infrared sky data and identify 1.5 million variable-object candidates. The pipeline processes each source in under 53 microseconds and could yield a full infrared variability catalog once follow-up work confirms the objects, with implications for time-domain astronomy; Paz later won a top national science prize for the achievement.













