Behzad Vahedi

About Me

I am a software engineer with a strong geospatial, GeoML, and GeoAI background. I currently work at X, the moonshot factory, on Bellwether—building machine learning systems that help predict and understand changes across the Earth.

I earned my Ph.D. in Geography from the University of Colorado Boulder in 2024, where my research focused on scale and uncertainty in geospatial AI, including deep learning for sea ice classification from radar imagery. Earlier, I completed an M.A. in Geography at UC Santa Barbara and degrees in GIS and Geomatics Engineering at K.N. Toosi University of Technology. My work has appeared in venues such as Nature Communications, IEEE JSTARS, and NeurIPS workshops.

Experience

Software Engineer, Bellwether · X, the moonshot factory
Sep. 2024 – Present · Boulder, CO

• Building GeoAI systems for Earth observation and environmental prediction on the Bellwether team
• Co-authored work on scalable geospatial data generation with Google DeepMind’s AlphaEarth Foundations model (arXiv; NeurIPS 2025 Climate Change AI workshop)

Ph.D. Resident · X, the moonshot factory
Aug. 2023 – Feb. 2024 · Mountain View, CA

• Developed AI models for forecasting extreme climate events
• Built CI/CD pipelines for automated testing and deployment across environments

Remote Sensing Data Science Intern · Earth Lab / CIRES, University of Colorado Boulder
May 2023 – Aug. 2023 · Boulder, CO

• Extended FIREDPy for near-real-time fire event perimeter delineation via remote sensing data fusion

Graduate Research & Teaching Assistant · University of Colorado Boulder
Aug. 2020 – Oct. 2024 · Boulder, CO

• Led research on sea ice classification/segmentation and spatiotemporal COVID-19 forecasting
• Designed labs and TA’d for Spatial Machine Learning and Data Science and related courses

Skills

• Focus: Machine learning, deep learning, GeoAI / GeoML, computer vision for Earth observation, MLOps
• Programming: Python, SQL, Java, R
• ML / DL: PyTorch, TensorFlow, Scikit-Learn, Hugging Face, Weights & Biases
• Geospatial: GDAL, GeoPandas, Rasterio, Google Earth Engine, ArcGIS, QGIS
• Infrastructure: GCP, Docker, Git, GitHub Actions, CI/CD

News

• [Dec. 2025] Our paper on scalable geospatial data generation with AlphaEarth Foundations was presented at the NeurIPS 2025 Climate Change AI workshop.
• [Aug. 2025] Preprint: Scalable Geospatial Data Generation Using AlphaEarth Foundations Model.
• [Jul. 2025] Preprint: Investigating the Effect of Spatial Context on Multi-Task Sea Ice Segmentation.
• [Apr. 2025] Our AutoICE sea ice charting paper is published in Remote Sensing Applications: Society and Environment.
• [Sep. 2024] Started as a Software Engineer on Bellwether at X, the moonshot factory.
• [2024] Completed my Ph.D. in Geography at the University of Colorado Boulder.
• [Jun. 2024] Our paper on Partial Label Learning for sea ice classification was accepted in IEEE JSTARS.
• [Feb. 2024] Finished my AI Residency at X / Google X (Bellwether).

Education

• Ph.D., Geography · University of Colorado Boulder · 2020–2024
  Dissertation: Incorporating Scale and Uncertainty in Geospatial Artificial Intelligence
• M.A., Geography · University of California, Santa Barbara
• M.Sc., GIS Engineering · K.N. Toosi University of Technology
• B.Sc., Geomatics Engineering · K.N. Toosi University of Technology

Selected Publications

A selection of recent and representative work. Full list on Google Scholar.

Presentations & Workshops

• [NeurIPS 2025] Scalable Geospatial Data Generation Using AlphaEarth Foundations Model — Tackling Climate Change with Machine Learning workshop, San Diego
• [AGU 2022] Sea Ice Type Classification using Deep Convolutional Networks and Partial Label Learning — Invited Talk
• [AGU 2021] A Comparison Of Classic Deep Learning Architectures For Sea Ice Classification From SAR — Outstanding Student Presentation Award (OSPA)

Earlier conference presentations are listed on Google Scholar and my LinkedIn.

Professional Service

Journal reviewer: IEEE JSTARS · PLOS One · Scientific Reports · Annals of GIS · International Journal of Digital Earth · Atmospheric Research

Powered by Jekyll and Minimal Light theme.