GraphBeat

GraphBeat is a music discovery engine that maps songs into a Neo4j knowledge graph based on “Musical DNA” — objective attributes like BPM, energy, and mood pulled from the Soundcharts and Last.fm APIs. Given 2–3 seed songs, it traverses the graph to surface “bridge” tracks that connect them across genres and eras, then uses Llama 3.3 to generate a personalized explanation of why each bridge works.

Built at UT Austin’s Women in CS Hackathon (February 2026) alongside Ashi Sharma and Maggie Li. My contribution centered on the app architecture and the graph layer specifically: designing the Neo4j schema through the neomodel OGM to enforce data integrity across a multi-node graph of songs, traits, and relationships, alongside the Django/LangChain pipeline connecting graph traversal to the Llama 3.3 explanation step.

Stack: Neo4j AuraDB, neomodel, Django REST Framework, LangChain, Llama 3.3 (via Groq), React, Tailwind CSS