Compress · Features
Everything it takes to make data small.
Neural compression that makes huge files small, with the error you choose.
ERA5 · 2 m temperatureBound 0.01 K · verified
Neural codecs
Codecs trained on your kind of data.
- Pretrained for gridded, image, audio and video data
- Fine-tuned on your own archive
- Lossless and lossy modes
Error bounds
Set the error per variable; we guarantee it.
- Absolute or relative bounds
- Per variable and per region
- Verification report with every file
Scientific formats
NetCDF, GRIB, Zarr, GeoTIFF and more.
- Drop-in for existing pipelines
- Chunked for cloud reads
- Read a slice without decoding the file
Edge & streaming
Compress at the sensor, before the link.
- Runs on small devices
- Made for satellite and cellular links
- Streams, not just files
SDK & CLI
One line in Python, a command in the shell.
- Python, C++ and Rust
- Command line for batch jobs
- Plugins for xarray and GDAL
Cloud, on-premises or air-gapped
Runs where your data already lives.
- Your cloud account or ours
- On-premises clusters and HPC
- Air-gapped installs
Questions.
Which formats does it read?
Placeholder: NetCDF, GRIB, Zarr, HDF5, GeoTIFF and common image, audio and video formats, with plugins for xarray and GDAL.
Do I have to decompress a whole file to read part of it?
No. Files are chunked, so you can read a slice, a region or a single variable straight from cloud storage.
Where does it run?
In your cloud account, on-premises, air-gapped or on small devices at the edge. Your data never has to leave your control.
How is it priced?
Placeholder: on the volume you compress and how you deploy it. Book a demo and we’ll put a plan together with you.
We’ll cover
- Compress run on a sample of your data
- Size, speed and error compared with what you use now
- Fitting it into your pipeline
- Deployment: cloud, on-premises or edge