Tibbe-AG

# The Problem

Traditional LLMs often generate hallucinated medical responses without reliable references, making them unsuitable for biomedical use cases.

# The Solution

Developed a GraphRAG architecture that retrieves structured biomedical knowledge from Neo4j and grounds every AI-generated response with relevant evidence before presenting it to the user.

# My Role

• Designed the overall system architecture.
• Built the FastAPI backend APIs.
• Developed GraphRAG retrieval pipeline.
• Integrated Groq LLM APIs.
• Designed and implemented the React frontend.
• Modeled the biomedical knowledge graph in Neo4j.

# Technology