MongoDB: We utilized MongoDB Atlas to handle massive IoT sensor throughput, pushing it far beyond a standard database by leveraging its Time-Series collections. Our FastAPI backend continuously piped high-frequency, second-by-second biometric vectors directly into Atlas. This architecture allowed us to log a distinct document per user, per song, and per second. Each document contains raw metrics like BPM, HRV, Presage facial expression confidence, and Muse alpha/beta waves alongside backend-derived states for mathematical arousal and valence. This frame-by-frame granularity expertly handles writes made by our streaming sensors and is exactly what differentiates Museic from legacy platforms like Spotify.
Additionally, we used MongoDB to store our pre-processed music library, which was seeded offline via a Python script prior to the event. This collection houses standard track metadata, Feed UI album graphics, and Gemini-generated JSON arrays that classify instruments, vocal textures, and specific moods. Finally, the database stores dynamic, authenticated user profiles keyed by their Auth0 user IDs. These profile documents house the user's saved reaction analyses, their personalized Gemini-generated music profiles, and the social connection graphs used to link them with users who share similar biometric taste in music.
Log in or sign up for Devpost to join the conversation.