Foundation Models & EO Embeddings
Generating dense, global embeddings from Sentinel-1 and Sentinel-2 at scale (Major TOM) — turning petabytes of imagery into a compact latent space that powers downstream AI.
GeoAI · Foundation Models · Earth Observation
Turning satellite signals into insights
01 — About
I lead GeoAI algorithm development for Earth Observation at CloudFerro — building the models and pipelines that turn petabytes of satellite data into ready-to-use AI: global embeddings and foundation models, semantic search, anomaly detection and analysis-ready datasets. My background is in remote sensing and machine learning, and my focus now is making AI-for-EO scalable, reliable and reproducible on cloud infrastructure.
02 — Research
Generating dense, global embeddings from Sentinel-1 and Sentinel-2 at scale (Major TOM) — turning petabytes of imagery into a compact latent space that powers downstream AI.
Text-to-image and image-to-image search over massive EO archives using multimodal embeddings — finding what matters in satellite data without hand-crafted labels.
Building on EO embeddings to detect change and anomalies for early-warning monitoring — configurable, reliable indicators over large areas of interest.
Curating and generating GPU-optimized training datasets and analysis-ready data cubes (ZARR) — lowering the barrier to entry for AI-for-EO research and production.
Production-grade processors on cloud infrastructure — cloud masking, Sentinel-1 terrain correction (RTC/ARD) and super-resolution — the reliable foundation under every AI service.
Exploring natural-language, tool-using agents (via MCP) that orchestrate multi-step EO analysis — an intelligent co-pilot for Earth Observation workflows.
03 — Path
CloudFerro S.A.
Leading GeoAI & ML algorithm development over large-scale open satellite archives.
CloudFerro S.A.
Foundation models, embeddings and EO pipelines on cloud infrastructure.
University of Warsaw
NCN Preludium grant — spruce forest damage from Sentinel-2 time series using ML.
Institute of Geodesy and Cartography (IGiK)
ML for vegetation, grasslands and crop monitoring across national and EU projects.
Institute of Geodesy and Cartography (IGiK)
Institute of Geodesy and Cartography (IGiK)
University of Warsaw
University of Warsaw
University of Warsaw
04 — Blog
Notes on GeoAI, foundation models and Earth Observation are on the way — first posts landing shortly.
05 — Publications
06 — Contact
Open to collaborations on Earth Observation, remote sensing and applied ML. The fastest way to reach me: