Nyas is a lightweight and efficient vector database engine designed for similarity search of embeddings. It supports multiple distance metrics (L2, Cosine, Dot), hybrid CPU/GPU execution, and flexible storage backends.