In a company-wide hackathon, the team at Memgraph developed a Spotify song recommendation engine using Kafka and Memgraph, leveraging an open dataset of Spotify playlists. The application architecture involves a Vue.js web interface for creating playlists and suggesting tracks, with a backend powered by Python Flask that queries Memgraph using custom MAGE algorithms for recommendations. The data model uses a graph structure where playlists are connected to tracks, and songs are recommended based on factors like playlist frequency, proximity, and influence. The backend server offers various REST endpoints for playlist and track management, while the recommendation algorithms suggest tracks by traversing the graph and evaluating song similarities. The project, which is open for public exploration on GitHub, highlights both the potential and areas for improvement in creating scalable music recommendation systems.