CV
Education
University of Wisconsin-Madison
Ph.D. in Geography/Environment (GeoAI, ML, and LLMs for Spatial & Human Data), Exp. 05/2028
M.S. in Cartography/Geographical Information System, GPA: 3.91/4.0 05/2025
Double M.S. in Computer Science (expected 2026) 09/2023 - Exp. 05/2028
University of Washington-Seattle
Bachelor of Art Geography: Data Science, GPA: 3.81/4.0 09/2019 - 03/2023
Research Experience
Lead Developer | OSU ICICLE Program - GNN Food Flow January 2025 – Present
- Built scalable data and training pipelines for GNN-based prediction of 10M+ food flows, leveraging distributed computing on HPC clusters. Prototyped models with PyTorch and optimized performance metrics (AUC 0.82, R² 0.36).
- Using inference results in flow volume regression across food categories, with developing open-source outputs for national-scale infrastructure resilience studies.
Lead Researcher | GeoAnalystBench Project October 2024 – Present
- Developed and deployed a benchmark of 50+ tasks to evaluate LLMs (Transformer-based architectures) on reasoning and code generation; designed data pipeline and evaluation framework (CodeBLEU, workflow metrics).
- Designed the evaluation framework using metrics like CodeBLEU and workflow difference to measure alignment between LLM and human reasoning; results showed proprietary LLMs outperformed open-source models on task structure and semantic flow.
- Categorized tasks into applied spatial analysis types (e.g., “Determining Relationships”, “Predictive Modeling”), revealing that LLM performance varies significantly across GIS domains.
Researcher | Urban Wildlife & Human Mobility September 2024 – Present
- Led a study analyzing over 5,000 coyote sightings with multi-scale human mobility data (SafeGraph) in Los Angeles County during and after the COVID-19 pandemic.
- Framed an ambiguous ecological question (coyote-human interactions post-Covid) into ML tasks; applied SEM and GWR for spatiotemporal modeling, integrating human mobility data.
- Discovered significant inflow/outflow effects on urban coyote presence, demonstrating behavioral plasticity to human movement.
Research Assistant | OSU ICICLE Program - ScienceAgentBench April 2024 - October 2024
- Designed and annotated 25+ complex GIS and spatial reasoning tasks to benchmark large language models on geospatial intelligence.
- Contributed to a 15% improvement in GIS-specific LLM task accuracy through the design of reproducible workflows and evaluation scripts.
- Developed a taxonomy of spatial agent task types to improve the scientific rigor of GIS AI benchmarks.
Conferences Talks
Scalable Inter-County Food Flow Prediction Using Graph Neural Network
Talk at ACM 2025 SigSpatial, Minneapolis, Minnesota, United States
Automating Geospatial Analysis Workflows Using ChatGPT-4
Talk at ACM 2024 SigSpatial, Atlanta, Georgia, United States
Enhancing Wildlife Classification Accuracy in Camera Trap Images Using GIS-Enhanced Federated Machine Learning Approaches
Talk at AAG 2024 Annual Meeting, Honolulu, Hawaii, United States
The Ethics of AI-Generated Maps: DALL·E 2 and AI’s Implications for Cartography
Talk at The 12th International Conference on GIScience, Leeds, United Kingdom, Leeds, United Kingdom
Grants & Awards
- NSF Student and Early Career Scholarships, 08/2025
- Trewartha Research Awards, University of Wisconsin-Madison, 03/2025
- I-Guide Summer School Travel Awards, 08/2024
- ESRI User Conference Student Assistantship, 07/2024
- Trewartha Conference Travel Awards, University of Wisconsin-Madison, 03/2024
- Dean’s List, University of Washington-Seattle, 01/2020 - 03/2023
Teaching
Skills
- Programming: Python (Proficient), SQL, R, Java, JS
- ML/AI Frameworks: PyTorch, HuggingFace Transformers
- Specialized: LLMs, GNNs, CNNs, data & training pipelines, distributed ML (HPC, AWS/GCP basics)
- Other: ArcGIS/QGIS, visualization (Tableau, Illustrator, Figma), Docker, Git
Service and Leadership
- Member, GeoDS Lab, UW–Madison
- Reviewer, ACM SigSpatial Workshop, IJGIS