I plan to build and launch a CubeSat with a LLM onboard. This first iteration aims to test the feasibility of running LLMs in space. While traditional spacecraft use onboard computers for data processing and control, I'll use a Raspberry Pi running a local LLM model, likely Google's Gemma.
Communication flow: Ground Station → CubeSat (Raspberry Pi) → LLM Processing → Ground Station
Many unknowns remain, including costs (hardware and launch), power consumption, and technical challenges. This blog will document my learning process and challenges encountered. Follow my quick updates on X (Twitter) @akasakahills.