2026 · Runner-up · HAL Aerothon '26
RagVault
Fully offline, privacy-first enterprise RAG system. Document intelligence with zero data leaving the network.
What I built
- Built a fully offline, privacy-first enterprise RAG system: no external API calls, no cloud, no data leaving the network. Runner-up at HAL Aerothon ’26.
- Quantized local LLM inference via Llama.cpp, vector search with ChromaDB and orchestration with LangChain behind a FastAPI service.
- Multi-format document ingestion and department-scoped access control so teams only query their own documents.
Stack
PythonFastAPILangChainLlama.cppChromaDB