Lemonade by AMD: a fast and open source local LLM server using GPU and NPU

AMD hardware is often under-leveraged for local LLMs compared to NVIDIA. Create a standardized middleware or runtime wrapper that abstracts the complexity of NPU-GPU hybrid acceleration.

Suggested repo: luma-serve

"Unlock your NPU: The first truly fast local LLM server for AMD hardware."

Estimated effort: 80h