KnowledgeOS.
Structured knowledge graph for AI-augmented retrieval and reasoning
Overview
KnowledgeOS ingests documents, extracts entities and relations via LLM pipelines, and stores them in a property graph (nodes typed as Person, Organization, Concept, Event; edges typed with relation labels). At query time it combines graph traversal (shortest path, neighborhood expansion) with vector similarity search, returning ranked passages enriched with entity context. The system tracks provenance — every fact links back to its source document and extraction timestamp. Used as the memory backbone for AURA.
Technology stack
TypeScriptNeo4j
#knowledge-graph#rag
Status
In Development
Published · Jun 2025
Updated · Jun 2026
Links
Public portfolio repository pending
Repository pending — real references to be added when available.
Related Lab