Project Citadel: Building a Fully Sovereign Dual-Engine GraphRAG on Consumer Hardware
A developer has created a personal, fully sovereign AI system called Project Citadel to manage local documents. This system utilizes a dual-engine approach combining a graph database (Neo4j) and a vector database (pgvector) with the Gemma 4:31b model. The project emphasizes local data control, avoiding third-party AI services for sensitive personal information. The architecture includes a hybrid compute setup with a powerful CPU and an eGPU for efficient processing. AI
IMPACT Enables local, sovereign AI solutions for personal data management, bypassing large cloud providers.