Spotify Playlist Generator
Built a Spotify-style playlist builder with search and ML-driven recommendations that update based on the user's evolving playlist.
Overview
This project simulates a Spotify-like experience where users can search tracks, build playlists, and receive personalized recommendations from that playlist. I implemented the core recommendation logic using similarity-based ML techniques, turning playlist context into ranked suggestions that adapt as tracks are added or removed. On the product side, I focused on making the workflow feel real: fast search, clean UI states, and recommendation panels designed for iteration. On the engineering side, I structured the system so the recommendation pipeline and search layer are modular and testable, enabling quick experimentation with feature weighting, ranking logic, and evaluation metrics.
Highlights
- Playlist creation UX with real-time updates (add/remove tracks, responsive UI states)
- Search engine for tracks with ranked results and query handling for fast discovery
- ML-based recommendation pipeline (playlist-to-track similarity + ranking) that adapts as the playlist changes
- Modular architecture for experimentation (feature weighting, ranking tweaks, and evaluation metrics)