khushpatel2002

RelationalDB to GraphRag Transformation

Problem Statement:

Build an automated, production-ready system that transforms structured relational data into a semantically enriched knowledge graph to enable relationship-aware, multi-hop, and semantic retrieval.

---

Requirements

khushpatel2002

Build Agentic Graph RAG as a Service

Problem Statement

Build an extensible, production-grade platform that unifies knowledge from multiple sources into an intelligent retrieval system. The platform must:

1. Automatically construct knowledge graphs from unstructured documents using LLM-generated ontologies and OpenAI embeddings, with a visual editor for ontology refinement and retrieval testing.
2. Provide a unified retrieval server that combines three complementary search methods—vector similarity search using OpenAI embeddings for semantic matching, graph traversal for relationship-based queries, and logical filtering for metadata/attribute constraints—all orchestrated by autonomous AI agents that dynamically determine optimal retrieval strategies based on query complexity, enabling users to extract insights through natural language queries that seamlessly blend semantic understanding, relational reasoning, and precise filtering in a single, cohesive system.

Requirements