Out-of-core LLM inference engine written from scratch in Rust

**Brief Summary (2‑3 sentences)** The project is a Rust‑native, out‑of‑core inference engine for large language models (LLMs) that runs directly from disk, avoi