Back to Projects
Knowledge Graph & RAG

Babcock University Knowledge Graph Ecosystem

A production-grade, full-stack AI platform that integrates Large Language Models (LLMs) with a structured Knowledge Graph (Neo4j) to provide factual, high-precision answers regarding Babcock University's academic and administrative data.

Executive Summary

This ecosystem serves as the central fact-checking hub for Babcock University data, enabling researchers and students to query complex academic relationships that are impossible to navigate via traditional SQL databases alone. It features a robust FastAPI backend, a high-performance React dashboard, and a published Python SDK for external developer integration.

Featured Library: gdg-bu-kg

A professional-grade Python package developed to allow developers to interact securely with the Knowledge Graph. Key features include:

  • Hybrid Authentication: Implements a "One-Time Exchange" mechanism where permanent API keys are traded for short-lived Session JWTs, reducing query latency by ~500ms.
  • State Management: Fully Asyncio-compatible client built on Pydantic v2 models for strict type safety.
  • Security Architecture: Internal attribute mangling and backend proxying to prevent data exposure.
  • AI Integration: Wrapper for conversational querying with configurable system prompts and temperature settings.

Key Engineering Achievements

  • Systems Optimization: Optimized the authentication flow to use stateless JWT validation, significantly improving response times for high-concurrency SDK usage.
  • Dialect-Aware Database Engine: Engineered a robust database connection pool in SQLAlchemy with automatic PostgreSQL/SQLite failover and connection recycling.
  • Command Center UI: Designed and implemented a real-time monitoring dashboard providing instant visibility into query volume, system reliability, and active user entities.
  • Audit-Driven Analytics: Built a comprehensive audit logging system that tracks system performance, query patterns, and failure rates across 30-day windows.

Impact

The platform serves as the central fact-checking hub for Babcock University data, enabling researchers and students to query complex academic relationships that are impossible to navigate via traditional SQL databases alone.