Kelsey Doerksen

Kelsey Doerksen

Postdoctoral Researcher

Hi, I'm Kelsey ๐Ÿ‘‹

Iโ€™m currently a postdoctoral researcher in geospatial AI, holding a joint appointment at Arizona State University under Hannah Kerner and the University of Cape Town at the Climate Risk Lab with Chris Trisos. I completed my PhD at the University of Oxford as part of OATML, supervised by Professor Yarin Gal in the Autonomous Intelligent Machines and Systems Doctoral Training Program.

My research is rooted in building AI tools for Earth Sciences that support scientific understanding. I have worked with NASA, ESA, and UNICEF on building AI tools to support climate modeling, disaster preparedness and humanitarian capacity building.

Prior to my PhD, I was a Space Systems Engineer in Satellite Operations at Planet, operating the world's largest EO constellation! I completed my Masters in Electrical and Computer Engineering and the University of Western Ontario and my Bachelors in Aerospace Engineering: Space Systems Design at Carleton University.

Code & Projects ๐Ÿ’ป

Open-source tools, datasets, and research code. More on GitHub.

Tools & datasets

airPy

Generate AI-ready datasets for air quality and Earth observation studies using Google Earth Engine.

โ˜… 90 ยท 8 forks repo โ†—
earthshift

Benchmarking the robustness of geospatial foundation models to realistic distribution shifts in Earth observation.

โ˜… 42 ยท 4 forks repo โ†—

Research code

giga-connectivity

Predicting school connectivity around the world from EO data. ESA ฮฆ-lab ร— UNICEF Giga research fellowship.

repo โ†—
giga-mobility

Mobility data analysis to support AI mapping of schools for the UNICEF-ITU Giga project.

repo โ†—
sudsaq

Deep learning with uncertainty quantification for physical-model bias of surface ozone.

repo โ†—
landslide-forecast-nepal

Forecasting landslide probability in Nepal with machine learning and EO data.

repo โ†—

News ๐Ÿ“ฃ

  • October 2026 My paper, AI-powered sub-seasonal forecasting of monsoon-triggered landslide impacts in Nepal has been published in npj Natural Hazards!
  • October 2026 Our work is accepted to AGU! I will be presenting two orals on AI-powered crop impacts and EarthShift
  • September 2026 I'll be at NYC Climate Week โ€” reach out if you'd like to meet up!
  • September 2026 I gave a talk at the TreeSense lab at University of Copenhagen!
  • September 2026 I presented at the TerraBytes Workshop at ECCV 2026!
  • August 2026 Our paper was accepted to the CV4Ecology workshop at ECCV 2026 - arXiv coming soon.
  • July 2026 I graduated my PhD from Oxford!
  • May 2026 New pre-print: EarthShift: a benchmark for measuring robustness to real-world distribution shifts in Earth observation.