Adam Stewart ’14 grew up collecting fossils along a creek near his childhood home in Tioga Center, New York, fascinated with paleontology. Today, he is a postdoctoral researcher at the Technical University of Munich in Germany, developing artificial intelligence tools that help scientists better understand the Earth through satellite imagery.
The path between those two points has been anything but straightforward. It has taken Stewart through computer programming, national laboratory research, graduate studies in artificial intelligence and the creation of a widely used open-source software platform called TorchGeo.
Despite these seeming zigs and zags, his work has followed a consistent thread tied firmly to that curiosity about the natural world that first drew him to earth science.
“I was always that kid who never stopped asking why, but I’ve also never stopped answering why. I’m driven to figure out the why if no one else knows it already,” Stewart said. “And that’s exactly what researchers do.”
Stewart’s fascination with science was nurtured by his experiences in Boy Scouts, including a camp on Cayuga Lake where he eventually became an instructor teaching natural science.
Although relatively few people from his hometown pursued higher education, Stewart’s parents encouraged him to attend college. He arrived at Cornell intending to study paleontology, but gradually discovered that his interests extended beyond fossils.
“Since I was in the first grade I would tell people, ‘I want to be a paleontologist. I want to study dinosaurs,’ and I never fully grew out of that phase,” Stewart said. Though working at Ithaca’s Paleontological Research Institution and taking courses in Cornell’s Department of Earth and Atmospheric Sciences did expose him to new scientific possibilities.
A turning point came when Stewart noticed a research poster in Snee Hall illustrating how satellite imagery could be used to track the movement of glaciers. The images captivated him. “This is what I want. I want something visual, something that’s just gorgeous, something beautiful,” he recalled thinking.
Stewart joined research led by professor Matthew Pritchard, studying glacier movement in the Russian High Arctic. During this time, Stewart worked alongside Andrew Melkonian, Ph.D. ’15, a doctoral student with extensive programming experience, sparking his own interest in computer science. Cornell’s flexible science of Earth systems curriculum allowed Stewart to create his own concentration in computational geophysics, combining earth science with mathematics, physics and programming.
After graduating, Stewart worked as a software engineer at Lockheed Martin before joining Argonne National Laboratory as a high-performance computing system administrator. There, helping researchers install and manage scientific software introduced him to open-source development, including work on Spack, a software management tool for scientific computing.
Stewart subsequently earned a doctorate in computer science at the University of Illinois Urbana-Champaign, where his research increasingly combined machine learning with Earth observation.
During an internship with Microsoft’s AI for Good lab, he began developing TorchGeo, a software library designed to make it easier for researchers to apply artificial intelligence to geographic and satellite data.
TorchGeo has since grown into an international collaborative project, attracting approximately 150 contributors and millions of downloads. Stewart now chairs its independent organization, which brings together researchers and developers from academia and industry.
At Munich, Stewart continues developing AI models capable of analyzing large quantities of Earth observation data. He also teaches, an activity he has enjoyed since his days working with the scouts, and is pursuing opportunities to become a professor.
Looking back, Stewart sees a career shaped less by a carefully constructed plan than by a willingness to explore unexpected opportunities. From paleontology to geophysics to artificial intelligence, his interests have evolved considerably. But the underlying motivation remains much the same: understanding the Earth and finding better ways to protect it.