music magal

MusicMagal is a group music recommendation system designed to solve a very real problem: creating a playlist that everyone in the group actually likes.

It works by pulling user data from Last.fm for a group of people, then modeling their musical preferences using an alternating least squares (ALS) recommender and item2vec embeddings.

The model produces a playlist of n songs optimized to satisfy m users. The final playlist is automatically created and saved to Spotify via its Web API — no manual curation needed.

This project combined collaborative filtering techniques with NLP-inspired embedding strategies to reflect not just what people like individually, but what resonates across a group. It was a fun technical challenge with a clear human-centered use case.

You can explore the full repo here or read the full writeup on Hackernoon.