Wuhan University
Bachelor’s · Hydrology & Water Resources Science
2010–2014
Applied AI, remote sensing & computer vision
Yawen Shen沈亚文
I work on applied AI in Corteva Agriscience’s Remote Sensing and Computer Vision team. I’ve been applying AI to scientific problems since my PhD, and I also build AI-agent workflows for research and software development.
About
I’m part of the Remote Sensing and Computer Vision team at Corteva Agriscience, where our work spans imaging scales from microscopes to satellites. My work draws on computer vision, multimodal learning, and geospatial data to address agricultural problems.
AI has been central to my work since my PhD at the University of Virginia, where I applied machine learning to climate change and urban flooding. My research included training street-scale flood prediction models with physical simulations and using imagery to study flooding. I’ve continued applying AI throughout my industry roles.
Alongside my professional work, I develop AI-agent workflows for software development, research, and content preparation. These projects explore practical ways to divide tasks between agents, check their results, and keep human review in the process.
Education and Career timeline
Where I’ve studied and worked, from 2010 to today.
Georgia Tech
MS · Hydrology & Water Resources Science
2014–2015
University of Virginia
PhD · Engineering Systems & Environment
2016–2020
Arcadis
Water Resources Consultant
2020–2021
One Concern
Data Scientist
2021–2023
Corteva Agriscience
Applied AI · Remote Sensing & Computer Vision
2023–present
Corteva Agriscience
Applied AI · Remote Sensing & Computer Vision
2023–present
One Concern
Data Scientist
2021–2023
Arcadis
Water Resources Consultant
2020–2021
University of Virginia
PhD · Engineering Systems & Environment
2016–2020
Georgia Tech
MS · Hydrology & Water Resources Science
2014–2015
Wuhan University
Bachelor’s · Hydrology & Water Resources Science
2010–2014
Research landscape
AI methods connect my work across imaging, agriculture, climate, and flooding. Explore the topics below to see related papers and agent projects.
Hover over an item for a preview. Click or press Enter for the complete entry.Tap an item to preview it. Tap again or choose “View details” to open it.
Browse research by topic
Computer vision & multimodal AI
Images, sensing, and learned representations
Geospatial AI
Satellite observations and transfer across places
- AlphaEarth benchmarksPaper · 2026Evaluating AlphaEarth embeddings for crop yield prediction and mapping agricultural practices, including how well they transfer across locations.
- Satellite practice mappingPaper · 2026Using county-level statistics and satellite observations to map tillage practices without field-level labels for training.
Climate & flood modeling
Physical systems, machine learning, and resilience
- Flood prediction surrogatesPaper · 2020Training machine learning surrogates on detailed physics-based simulations to support real-time, street-scale flood prediction.
- Compound flood riskPaper · 2019Studying the combined effects of storm tide and heavy rainfall to understand flood risk and guide mitigation in coastal watersheds.
- Climate & infrastructurePaper · 2022Modeling how combined inland flooding and storm surge affect a coastal city’s transportation network under climate change scenarios.
AI agent workflows
Personal tools for research and software work
- Personal development teamPersonal project · Working locallyA Codex workflow that assigns focused software tasks to agents, preserves the work behind each change, and brings in an independent reviewer before delivery.
- AI trend trackingPersonal project · Working locallyA local research workflow for following AI developments, checking sources, and preparing concise briefings with links to the original material.
- Social Media ManagerPersonal project · In developmentA workflow I’m developing to research topics, prepare images, and draft Chinese-language posts, with human review before publication.
AI in practice
Current work
Corteva Agriscience
Our Remote Sensing and Computer Vision team works across scales, from microscope images to satellite observations. I apply computer vision and multimodal learning to agricultural problems within this team.
- Multimodal learning
- I combine imagery, remote-sensing observations, weather, and environmental data to study agricultural problems. This includes sensor fusion and aligning observations across space and time.
- Foundation models
- I work with geospatial foundation models and study how learned representations transfer to agricultural tasks. My research includes benchmarking AlphaEarth for crop-yield prediction and mapping agricultural practices.
- Project management & technical leadership
- I help turn scientific questions into applied AI projects: defining the problem, coordinating technical work, reviewing results, and communicating findings. This draws on my experience leading applied machine-learning projects and working with scientific and engineering collaborators.
Personal AI work
I also build agent workflows for tasks I use in my own research and development.
Personal development team
Working locallyA Codex workflow that assigns focused software tasks to agents, preserves the work behind each change, and brings in an independent reviewer before delivery.
AI trend tracking
Working locallyA local research workflow for following AI developments, checking sources, and preparing concise briefings with links to the original material.
Social Media Manager
In developmentA workflow I’m developing to research topics, prepare images, and draft Chinese-language posts, with human review before publication.
Selected publications
A selection of work across geospatial AI, flooding, and climate research.
- 2026Read paper
Geospatial foundation models
Harvesting AlphaEarth: Benchmarking the geospatial foundation model for agricultural downstream tasks
International Journal of Applied Earth Observation and Geoinformation, 149, 105258
Evaluating AlphaEarth embeddings for crop yield prediction and mapping agricultural practices, including how well they transfer across locations.
- 2026Read paper
Learning across scales
STaPL: Scale Transfer with Pseudo-Labelling for satellite-based mapping of agricultural practices
Remote Sensing of Environment, 343, 115500
Using county-level statistics and satellite observations to map tillage practices without field-level labels for training.
- 2024Read paper
Computer vision for flooding
Urban flood extent segmentation and evaluation from real-world surveillance camera images using deep convolutional neural network
Environmental Modelling & Software, 173, 105939
Extracting flood extent from camera images with deep learning, and examining the challenges of real storm conditions.
- 2022Read paper
Climate & infrastructure
Dynamic Modeling of Inland Flooding and Storm Surge on Coastal Cities under Climate Change Scenarios: Transportation Infrastructure Impacts in Norfolk, Virginia USA as a Case Study
Geosciences, 12(6), 224
Modeling how combined inland flooding and storm surge affect a coastal city’s transportation network under climate change scenarios.
- 2020Read paper
Machine learning & physical models
Training Machine Learning Surrogate Models From a High-Fidelity Physics-Based Model: Application for Real-Time Street-Scale Flood Prediction in an Urban Coastal Community
Water Resources Research, 56(10), e2019WR027038
Training machine learning surrogates on detailed physics-based simulations to support real-time, street-scale flood prediction.
- 2019Read paper
Compound flood risk
Flood risk assessment and increased resilience for coastal urban watersheds under the combined impact of storm tide and heavy rainfall
Journal of Hydrology, 579, 124159
Studying the combined effects of storm tide and heavy rainfall to understand flood risk and guide mitigation in coastal watersheds.
- 2019Read paper
Infrastructure deterioration
Condition State–Based Civil Infrastructure Deterioration Model on a Structure System Level
Journal of Infrastructure Systems, 25(1), 04018042
Modeling how interactions between structural components affect infrastructure deterioration, with uncertainty estimates to inform maintenance decisions.