E2
AI & ML interests
None defined yet.
Recent Activity
EmbedEarth
A world model for finding and computing anything about our planet.
EmbedEarth, or E2, is a platform for organizing the physical world into a searchable and computable system.
We are building a continuously expanding representation of Earth that makes it possible to find physical features across every city and region, understand where they are, connect them to surrounding context, and perform computation across them.
Our belief is simple: any feature of the planet should be searchable, and any geographic question should be computable.
What we work on
E2 is built around a world model that represents what exists, where it is, how it relates to its surroundings, and how it changes over time.
This includes infrastructure for:
- planetary-scale feature indexing
- geospatial search and retrieval
- places, buildings, roads, and addresses
- infrastructure and public assets
- natural and environmental features
- imagery and physical-world observations
- spatial and temporal computation
- routing, proximity, measurement, and aggregation
Our mission
Make the physical world searchable and computable.
E2 is designed to provide one system for working with geographic information across the planet.
A developer should be able to search for virtually any feature, in any city, and then immediately perform computation across the results.
Find every substation in a region.
Measure tree coverage around every school in a city.
Rank neighborhoods by access to parks.
Find buildings near flood zones.
Compare infrastructure across thousands of cities.
Compute routes, distances, densities, changes, intersections, proximity, rankings, and other geographic operations.
Why this matters
Today, information about the physical world is fragmented across maps, databases, imagery, government portals, sensors, and thousands of individual datasets.
Finding a particular feature can require discovering the right dataset, downloading it, cleaning it, joining it with other sources, and building custom spatial infrastructure before a single question can be answered.
E2 is working toward a different model.
One representation of the planet where physical features can be discovered through a common system and immediately used for computation.
The world model
E2 organizes geographic information around a common representation of Earth.
The goal is to make it possible to move seamlessly between:
- Search — find any feature of the planet
- Locate — determine precisely where it exists
- Relate — understand what surrounds or intersects it
- Measure — calculate distance, area, density, quantity, or change
- Compare — evaluate places, regions, and features against each other
- Compute — perform arbitrary geographic operations across the world
Every building, road, place, natural feature, infrastructure asset, observation, and region becomes part of the same computational environment.
On Hugging Face
This organization is where we publish selected datasets, geographic indexes, embeddings, derived features, and other resources produced through EmbedEarth.
These releases are part of a broader effort to make planetary-scale geographic information easier to discover, use, and compute over.
Vision
We believe the planet itself can become a computational layer.
Not a collection of disconnected maps and datasets, but a unified system where any feature can be found and any region can be computed over.
Every feature searchable. Every city computable.
EmbedEarth is building that system.