A team of two Purdue University graduate students earned a 4th-place finish in the 2026 Moonshot-Labs Analyst Jam for the Intelligence Community (MAJIC) Challenge, a month-long, national level collegiate competition focused on solving real-world intelligence and national security problems. Sponsored by the National Geospatial-Intelligence Agency (NGA), this year’s challenge centered on “Using AI and Geospatial Data Science: Modeling Conflict Risk and Physical Man-Made Obstructions”. Sixteen teams from universities across the country collaborated with subject matter experts to develop operationally relevant solutions.

Kendahl Klatt, M.S., and Owen Hodges, EIT, both graduate researchers in the Troy Lab within the Lyles School of Civil and Construction Engineering, competed with their project “Harnessing Machine Learning for the Identification of Vertical Obstructions in the United States”. Their work focused on developing an automated, scalable approach for detecting vertical obstructions (VOs) to enhance aviation safety, with case studies in airfields in the United States.

“In both military and aviation contexts, understanding the physical environment is essential to safety and mission success,” said Hodges. “Automating the detection of vertical obstructions improves situational awareness and ensures that critical data can be updated quickly and at scale.”

Using high-resolution lidar data and machine learning, the team built a pipeline to detect, classify, and estimate the height of potential obstructions to aviation. The model demonstrated strong detection performance and high accuracy. “Lidar data is powerful because it captures elevation, location, and multiple returns within a single dataset, giving us geometric and structural insight into features,” said Kendahl Klatt. “By training our model directly on those lidar returns, we streamlined a process that previously depended on stitching together multiple data sources, making it faster, cleaner, and scalable.”

By reducing reliance on manual surveying and enabling rapid, large-scale analysis, this work has direct implications for aviation safety, infrastructure monitoring, and national security applications. The approach also demonstrates the potential of scalable, open-source geospatial solutions to support timely updates to critical aviation databases.

As a result of their performance, the team was able to present their work to Senior Executive members of the intelligence community and has the opportunity to submit their research for publication through the NGA’s Tearline Project, further extending the real-world impact of their solution. While Klatt and Hodges’ primary research focus is on coastal and large-lake processes under Dr. Cary Troy, this project highlights the adaptability of their geospatial and machine learning expertise to broader national security applications.