In April 2025, I was able to share the idea of hosting your own language models at Interactive Media Arts, NYU Shanghai through a small workshop!

While on its own, setting up the solar server’s software/server system is quite simple! What was challenging was to figure it out at the University, where restricted internet access and proxy system’s setup are part of state mandates. Anyway after a tough few hours, this was the simple system I figured out on the wireless network of the 3D Printer (which didn’t have access to the internet!)

learnt a lot of bash commands! :P

The workshop included:

1. Accessing Raspberry Pi via SSH
2. Installing Docker and Ollama
3. Test Ollama with Tinyllama
4. Use the API endpoint from the Rapsberry Pi to control a p5.js sketch! :)

Sketch is hosted here!