Guest Seminar by Dr. Satish Kumar, EyeClimate Inc & Schmidt Science Fellow, Stanford Form

Speaker: Dr. Satish Kumar, EyeClimate Inc & Schmidt Science Fellow, Stanford

Date and Time: Nov 12, 2026 at 3:30 pm

Venue: CAS Seminar Room (VI-412)

Title: Foundation Models for Climate and Environmental Intelligence: Learning Across Sensors, Scales, and Extreme Events

Abstract: Understanding climate risks requires learning from observations that span different sensors, spatial scales, and time. At EyeClimate, our path toward multimodal foundation models began with a practical challenge: detecting and quantifying methane emissions from diverse satellite, airborne, and ground-based observations. Working across these modalities revealed both the value of complementary measurements and the challenges of aligning heterogeneous data, learning transferable representations, and generalizing across observation conditions.

This talk will trace how those experiences shaped our approach to building foundation models for broader climate and environmental applications. I will discuss how lessons from emissions monitoring inform the integration of satellite imagery and atmospheric fields, moving toward models that learn across sensing modalities and tasks. A central focus will be extending these capabilities to extreme-event detection and rainfall prediction, where rare, high-impact events challenge models trained predominantly on typical conditions.

Drawing on EyeClimate's development experience and a case study of extreme rainfall in the Texas Hill Country, I will examine the opportunities and limitations of multimodal learning, including data fusion, limited ground truth, and physically meaningful validation. The talk will conclude with opportunities to connect atmospheric science and AI research to develop reliable, observation-informed tools for climate adaptation and mitigation.

Speaker Bio: Satish Kumar is the Schmidt Science Fellow at Stanford University and CEO of EyeClimate Inc., with more than 13 years of research experience, including eight years at the intersection of climate science and artificial intelligence. His work focuses on multimodal foundation models for Earth observation, integrating satellite imagery, atmospheric variables, and ground-based measurements to support extreme-event detection, rainfall prediction, and environmental monitoring. This research spans collaborations with Stanford University, Argonne National Laboratory, Oxford University, and the University of Cambridge.

Satish's foundation-model research builds on his experience developing AI for greenhouse gas detection and wildlife monitoring in the Maasai Mara and Serengeti. Working across hyperspectral, multispectral, and aerial imagery shaped his approach to learning from heterogeneous observations and transferring knowledge across sensors, scales, and applications. At EyeClimate, he leads the core research and development of these capabilities, connecting advances in AI with practical challenges in climate adaptation, emissions monitoring, and conservation.

Satish holds a Ph.D. from the University of California, Santa Barbara, with a specialization in computer vision and machine learning. His research has appeared at leading venues including CVPR, NeurIPS, BMVC, and WACV, and he has delivered talks on AI for climate and wildlife conservation at international conferences and conservation organizations. He also brings five years of operating systems development experience from Samsung R&D in South Korea and India.