While most high school students spent their summer enjoying active recreation, teenager Matteo Paz from Pasadena decided to take on the challenge of NASA’s boundless data sets. The result of his work impressed even the professional community: the young man developed an artificial intelligence system that detected 1.5 million previously unknown space objects.
Matteo Paz with California Institute of Technology President Thomas Rosenbaum at the Regeneron Science Talent Search awards ceremony on March 11, 2025. Photo: Science Talent Search
His journey began in 2022 at the California Institute of Technology (Caltech) Academy. Under the guidance of his mentor Davy Kirkpatrick, Matteo gained access to the archives of the NEOWISE infrared telescope. This device scanned the sky for years, accumulating more than 200 billion records of thermal signatures in deep space.
From manual labor to a global algorithm
Starry sky. Illustrative photo: Unsplash
The initial task was modest: to analyze a small section of the sky manually in order to find several variable stars. However, Matteo, with his deep knowledge of mathematics and programming, decided to automate the process. Instead of relying on ready-made solutions such as ChatGPT, he created his own unique algorithm — VARnet.
“We were dealing with a table containing 200 billion rows accumulated over decades. Matteo wanted to analyze the entire sky at once, and I supported this ambitious idea,” recalls Kirkpatrick.
VARnet is a highly effective model for analyzing time series. It breaks down huge amounts of data into pieces and looks for distinctive infrared “fingerprints” in them. Thanks to this, the system was able to identify light pulsations belonging to binary stars, distant quasars, and even black holes that had previously gone unnoticed.
Practical future
Matteo’s work proved so significant that its results were published in the prestigious scientific journal The Astronomical Journal. Today, Caltech uses VARnet to study complex star systems, and the inventor himself received the grand prize in the prestigious Regeneron Science Talent Search competition — a scholarship worth $250,000.
The range of applications for VARnet extends far beyond astronomy. According to the author, the algorithm can be adapted for any data that changes over time:
Ecology: monitoring the spread of forest fires and pollution levels.
Finance: analyzing stock market fluctuations.
Medicine: tracking changes in patients’ conditions.
For Matteo, the environmental aspect is personal — his family was once evacuated due to a forest fire, so he strives to use his intellect to create early warning systems for disasters. This young man’s story proves that modern science has become democratic, and that all you need for a great discovery is a laptop, open data, and the courage to think globally.
Earlier, we reported on how Ukrainian schoolchildren’s experiments would be sent into space.
According to Futura