Neo Zhang
Code owner for llama.cpp's SYCL backend, which is designed primarily for Intel GPUs
- Code owner for the SYCL backend (ggml-org/ggml-sycl), llama.cpp project2
Neo Zhang, who publishes on GitHub as arthw, is the code owner for llama.cpp's SYCL backend, according to the project's CODEOWNERS file. The backend's documentation says it is designed primarily for Intel GPUs and lists supported Intel GPUs from integrated graphics to Arc and data center cards. Zhang's 2026 commits add operators, data types and FlashAttention support to the backend, improve its memory handling, and fix issues on specific Intel GPUs.1234
Work
llama.cpp SYCL backendOpen-source code
CODEOWNERS lists arthw as the ggml-org/ggml-sycl team, which owns ggml/src/ggml-sycl. Commits authored from July to October 2026 add sparse FlashAttention, more operators and quantized types, host-pinned and peer-to-peer memory support, and fixes for specific Intel GPUs, and update the backend's setup guide. The repository is MIT-licensed.245
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.