EPS Colloquium – Alexis Boukouvalas, Google Deepmind

AlphaEarth

We introduce AlphaEarth Foundations, a geospatial foundation model designed to overcome the challenge of mapping the planet when high-quality labels are scarce. While satellite data is abundant, translating it into accurate maps usually requires labor-intensive manual labeling or specialized models. AlphaEarth Foundations solves this by unifying spatial, temporal, and measurement contexts into a single, universal embedding field.

To be added to the EPS colloquium mailing list, please contact Caroline Carr at carolinecarr@fas.harvard.edu

Alexis Boukouvalas is a Research Engineering Team Lead at Google DeepMind. His work focuses on Bayesian statistics and non-parametric modeling. Alexis has published research on applications including single-cell bioinformatics, clustering algorithms, and spatiotemporal data modeling. His work includes optimisation of supply chains via surrogate modelling and interpretable machine learning methods.