Ishaan Dokania - Society for Science Skip to content

Ishaan Dokania

6th Grade, Willamette Valley Academy
Beaverton, OR

Eye in the Sky: From Pixels to Predictions for Lithium and Beyond

Lithium is essential for energy storage, but finding new sources to mine the mineral is difficult and costly. Ishaan trained a machine learning algorithm on satellite and mineral data and fed it additional geochemical markers to help it pinpoint new lithium sources. His model achieved 89 percent accuracy.

View Poster
2026 Thermo Fisher Scientific Junior Innovators Challenge Finalist - Ishaan Dokania: Eye in the Sky: From Pixels to Predictions for Lithium and Beyond
Eye in the Sky: From Pixels to Predictions for Lithium and Beyond Ishaan Dokania
View Poster

Project Background

While doing a Science Olympiad competition, Ishaan took on two topics: Remote sensing, and rocks and minerals. He began to wonder if he might combine his two interests to solve a real-world problem and learned about the lithium bottleneck. The world needs a lot of lithium for electronics and energy storage. “Traditional methods to find lithium is inefficient, wrecks the environment and is costly,” Ishaan says. He wondered if he could use remote sensing with satellite images to automatically identify lithium deposits.

Tactics and Results

Ishaan trained a machine learning model on satellite image data from Google Earth Engine, and data from the United State Geological Survey Mineral Resources Data System. He also identified features for his model that would predict lithium deposit locations. He tested several machine learning algorithms and identified several potential sources of noise in his data. For example, the models tended to have biased readings if they detected plants under heat stress, and would flag any hot region as containing lithium, making the model inaccurate. Ishaan added a test called Leave One Feature Out, to get rid of the noise. He also found that using different satellites for data could result in overlapping maps. Once his sources of noise were fixed, Ishaan’s model could detect lithium deposits with 89 percent accuracy.

2026 Thermo Fisher Scientific Junior Innovators Challenge Finalist - Ishaan Dokania
Ishaan Dokania

Beyond the Project

“I love playing badminton with my dad because we try to rally the birdie (keep on passing it back and forth) and we try to get a new record each time. My younger brother also competes with me in this and we each take turns playing with my dad,” Ishaan says. He would like to be an aeronautical engineer. “It is always fun for me to watch airplane videos on what parts of planes went wrong and different theories about how a plane might have done a crash or emergency landing,” he says.

2026 Thermo Fisher Scientific Junior Innovators Challenge Finalist - Ishaan Dokania
Ishaan Dokania