Radoslav Gerganov
Code owner for llama.cpp's RPC backend, which offloads computation to remote machines
Radoslav Gerganov is the code owner for llama.cpp's RPC backend, according to the project's CODEOWNERS file, which lists Gerganov as the sole member of the ggml-rpc team covering the backend and its server tool. The project's documentation says the RPC server exposes ggml devices on a remote host and that the RPC backend offloads computations to one or several of those servers. Gerganov's commits add multi-device support, compute-graph caching and reuse, a refactored transport, and asynchronous backend APIs to the RPC backend, along with request-handling changes in llama-server.123
Work
llama.cpp RPC backendOpen-source code
CODEOWNERS lists Gerganov as the ggml-org/ggml-rpc team, which owns ggml/src/ggml-rpc and tools/rpc; a September 2025 commit added Gerganov as RPC owner. Commits authored from September 2025 to August 2026 add multi-device support, graph caching and reuse, a transport refactor, and event and async backend APIs, plus llama-server changes such as returning HTTP 400 when a prompt exceeds the context length. The repository is MIT-licensed.1324
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.