product
one platform for scientific data
Store, govern, share, and compute on multidimensional data — built on the open standards our team authors.
get startedthe earthmover stack
One data model, top to bottom
Scientific data belongs in tensors, not tables — and the tools built for tables can't manage it. Earthmover is a full stack for multidimensional data: an open-source storage engine at the bottom, management and governance above it, and compute services on top — reached by people and by agents over the same interfaces. Each layer stands on its own, and they are designed to work together.
Humans
Agents
Tensor storage engine
Bring your own bucket
Or use storage we manage for you
generic S3-compatible open source
Icechunk
Tensor storage engine
A Zarr-compatible transactional storage engine, written in Rust. ACID transactions, git-like version history, best-in-class cloud I/O, and zero-copy virtualization of NetCDF, HDF5, GRIB, and TIFF archives. Apache 2.0, serverless, and yours to run anywhere.
explore icechunk →platform
Arraylake
Data management and governance
A managed data lake platform for your Icechunk repositories: a catalog that natively understands arrays, single sign-on and fine-grained access control, credential vending so users never handle cloud keys, and a marketplace for sharing data across organizations.
explore arraylake →platform
Compute
Scalable tensor compute services
Compute that runs next to your data and serves it over the protocols your tools already speak — map tiles, EDR, DAP2, openEO. Turn a repository into a data product without building or operating a delivery stack.
explore compute →marketplace
Data you don't have to ingest
The same platform that manages your own data can deliver someone else's. Subscribe to analysis-ready weather, climate, and Earth-system datasets from trusted providers and they appear as repositories in your organization — versioned, updated incrementally, and queryable immediately. No download, no ingestion pipeline to maintain.
how it's built
Design principles
Cloud native
Object storage as the primary data layer, storage and compute that scale independently. No filesystem assumptions anywhere in the stack.
Open standards
Built on Zarr, an Open Geospatial Consortium standard, and Icechunk, an Apache-2.0 project. No proprietary format stands between you and your data.
Bring your own bucket
Your data stays in your object storage, in your cloud account, under your policies — or use storage we manage for you.
Built for humans and agents
The same catalog, permissions, and APIs serve a scientist in a notebook, a production pipeline, and an AI agent exploring your archive.