Teen's AI Finds 1.5 Million Hidden Space Objects

A high school student just proved that the most important discoveries can come from looking at old information in completely new ways.

Matteo Paz from Pasadena built an artificial intelligence system that found 1.5 million space objects nobody had properly identified before. He didn't need a fancy telescope or expensive equipment. He just needed NASA's existing data and a smarter way to read it.

His system, called VARnet, reanalyzed infrared data from NASA's NEOWISE mission. The mission was designed to track asteroids, but it quietly collected something far more valuable along the way: signals from distant quasars, exploding stars, and eclipsing binaries that flicker and pulse across the universe.

The challenge was never missing information. It was overwhelming volume. NEOWISE collected nearly 200 billion individual detections over a decade, creating a 200 terabyte database that made manual analysis nearly impossible.

Paz started the project in 2022 as a summer experiment through Caltech's Planet Finder Academy. Working with astronomer Davy Kirkpatrick, what began as a six week research program quickly grew into something extraordinary.

His AI system learned to recognize patterns automatically, processing each space object in less than 53 microseconds. It achieved an accuracy score of 0.91, meaning it could reliably distinguish real cosmic changes from random noise.

The breakthrough earned Paz the top prize in the Regeneron Science Talent Search, along with $250,000. His peer-reviewed study was published in The Astronomical Journal in December 2024, and scientists are calling it a turning point in data analysis.

The Ripple Effect

The implications reach far beyond astronomy. Paz designed VARnet to analyze time series data, which means it tracks how things change over time. That same technique can monitor Earth's atmosphere, track pollution cycles, and detect climate patterns.

Paz himself explained that his model could "study atmospheric effects such as pollution" by analyzing repeating patterns shaped by seasons and daily cycles. Better environmental monitoring, earlier warnings, and more precise tracking of changes affecting human health could all benefit from this approach.

His discovery also reveals something profound about scientific progress. Vast archives of data sitting in databases around the world aren't finished stories. They're dynamic resources waiting for the right tools to unlock them.

The next major breakthrough may not require new missions or billion dollar telescopes. It might just need a fresh perspective and the courage to look differently at what's already there.