The L0 tax that nobody talks about
Sensor data does not arrive in formats anyone actually wants. It arrives as Sentinel SAFE packages with mission XML sidecars, as NITF/SICD complex products, as RINEX observation files, as SigMF I/Q captures, as raw CCSDS frames from a ground station. Every EO startup and every space operator pays a quiet tax to translate that into something downstream pipelines understand.
Level Zero is a normalisation fabric that pays that tax once, for everyone.
Three pipelines, one control plane
Three modular lanes :
- Imagery : optical and SAR scenes to COG, STAC items and collections, tiled map services.
- Signal & navigation : GNSS observations and SDR captures to RINEX 4, SigMF, decoded telemetry.
- Mission operations : orbit, attitude and tracking data to CCSDS OEM, OMM, AEM, TDM.
All three lanes sit on the same control plane with the same observability, the same versioned ingest contract, the same metadata envelope. From raw downlink to first cataloged STAC item, median 3.4 seconds.
Three point four seconds is not a benchmark. It is a customer experience.
Standards-first output, on purpose
Cloud Optimized GeoTIFF. STAC. OGC API Tiles. RINEX 4. SensorML. netCDF + CF where it earns its place. GeoParquet for vectors. SigMF for I/Q. CCSDS for ground-system handoff. The list is opinionated about what to ship in - and what to ship out.
The output formats are not a flex. They are the entire point. Standards-first is what makes Level Zero replaceable - a customer can adopt it without locking themselves into a proprietary ingest contract.
The engines we did not reinvent
GDAL for raster I/O. OTB and SNAP for EO preprocessing. ISCE3 and MintPy for SAR. SarPy for SICD interchange. RTKLIB for GNSS. GNU Radio plus gr-satellites for SDR. Orekit for orbit propagation. PgSTAC for catalog storage. The architectural choice is to compose these engines properly, not to rebuild what the open-source space and geospatial community has already gotten right.
Who it is for
EO startups and geospatial SaaS vendors who want reliable format conversion and STAC publication without owning the plumbing. Analytics teams who want their training corpora discoverable and reproducible. Ground and mission operations teams who need CCSDS-compliant exchange between their station and their partners. ML platform teams building training pipelines on top of STAC-labeled assets. The product page covers the full input class matrix.
Product · Level Zero
Sensor data, standardised
The full product page for Level Zero - the input classes, the three pipelines, the output formats, and the engines underneath.