Profile Page

Cédric M. John
Professor of Artificial Intelligence for Earth Sciences (as of September 15, 2026)
College of Earth, Ocean, and Atmospheric Sciences
Oregon State University

Lab website: John Lab

Github: John-Lab

Current Work in the John Lab

My research sits at the intersection of artificial intelligence and the Earth and planetary sciences, with a particular focus on computer vision and deep learning as engines for scientific discovery. I develop and apply foundation models, multimodal generative AI, and large-scale deep learning architectures to a range of geoscience data: core photographs, well logs, seismic volumes, satellite and airborne imagery, and the outputs of physics-based numerical simulators. A central thread is moving beyond bespoke, single-task models toward general-purpose representations of subsurface and surface data, so that the same backbone can support classification, segmentation, generation, and inversion tasks across very different Earth-science domains. Alongside this, my team and I build surrogate deep learning models that emulate computationally expensive geological simulations (notably forward stratigraphic models) turning multi-day numerical experiments into near-instantaneous inference and opening the door to large-ensemble uncertainty quantification, inverse problems, and AI-driven scenario exploration.

The applied side of this work targets some of the most pressing challenges facing our planet. For climate action, I use AI to interpret Earth observation data at scale, automate the analysis of environmental change, and improve forward stratigraphic models by combining data-driven models with physical priors and data assimilation. For the energy transition, the same computer-vision and generative-AI toolkit underpins automated subsurface characterization for carbon capture and storage, critical mineral exploration, hydrogen storage, and geothermal energy: domains where rapid, reproducible interpretation of satellite data,  core, log, and seismic data is a bottleneck to deployment. I am also increasingly interested in extending these methods to planetary sciences, where remote sensing of other worlds raises closely related machine-learning problems but with even sparser labels and more extreme distribution shift. Across all of these threads, the underlying ambition is the same: to make AI a first-class instrument of Earth and planetary science, as fundamental to the field as the microscope or the seismometer.

For more information on our work, visit the John Lab website.