A specialist with a narrow but well-defined focus: converting Turkish text into vector embeddings. It operates within a tight 512-token window, so it works best with shorter passages rather than long documents. As an open-weight model, it can be run locally without API dependencies.