Run AI applications
Deploy inference services as signed, hardware-isolated OCI workloads with explicit access to the accelerator and data interfaces they need. There are GPU, FPGA, AI engine, and many-core target profiles.
Parsimoni Platform
Secure infrastructure to run AI and mission applications in space, with hardware-backed isolation across spacecraft and edge targets.
The Parsimoni Platform brings the build, run, and deploy workflow of the container ecosystem to spacecraft payload software. Applications ship as OCI artifacts, run as hardware-isolated unikernels or Linux containers, and keep the same logical identity from a developer laptop to a spacecraft target.
The Platform has four components: Space CLI builds, Space Store publishes, Space CCSDS carries software across the space link, and SpaceOS is the on-board runtime that isolates and supervises workloads on the spacecraft. run.space provides the hosted scheduling and operations surface.
Choose a task
Run a workload locally
Install Space CLI, build a signed composition, and boot it in a development VM in about ten minutes.
Integrate a target
Choose a full-OS or daemon installation based on who owns the target’s boot path.
Understand the system
The Platform separates flight and payload software, isolates services, and connects applications to spacecraft data planes.
Build, run, and deploy
space build produces a signed composition without tying it to one computer.
space run boots it on the local development runner. space deploy prepares
it for a spacecraft or edge computer and applies that target’s hardware and
security settings. Application identity and policy remain explicit in both
paths.
Accelerated inference targets include the NVIDIA Jetson TX2i for CUDA and TensorRT, the Xilinx VCK190 for Vitis AI, and the Kalray Coolidge for many-core edge AI.
