📚 NornicDB: Graph and Vector Database with Built-in MCP

In short, this project tries to merge graph and vector databases into a single distributed, low-latency system. It features temporal MVCC and claims sub-millisecond performance for HNSW vector search, graph traversal, and writes. The interesting part is its drop-in compatibility. By supporting Neo4j's Bolt/Cypher and Qdrant's gRPC, it allows developers to switch backends without rewriting queries. On top of that, it adds AI-native features like managed embeddings, reranking, GPU acceleration, and a built-in MCP server. It is an ambitious take on what the next generation of AI infrastructure should look like.