On-device inference
Getting real LLM performance out of constrained hardware — from a Raspberry Pi to a GPU workstation.
Building sovereign AI, on real hardware
We don't have specific open roles listed right now — this page will be updated when we do. If you want to work on private, hardware-native AI across edge, on-prem, and automotive deployments, we'd still like to hear from you.
What we work on
RAG & command execution
Retrieval pipelines and a validated command layer that lets a model safely trigger real actions.
Hardware & systems
Mapping software to a wide, real hardware catalog — edge accelerators, mini PCs, on-prem servers, and automotive reference kits.
Want to introduce yourself?
Send your background and what you'd want to work on — we read everything.