Personify

Analyzed Spotify music data to understand which audio metrics — energy, valence, danceability, acousticness — most meaningfully differentiated listener types. We determined threshold divisions to produce roughly equal-sized personality buckets.

• Explored Spotify audio feature distributions across genre and mood
• Identified 4 primary metrics that best segmented listener behavior
• Designed threshold logic to balance personality distribution across users

Used early GPT models to brainstorm personality names and archetypes that felt evocative rather than clinical — we wanted names that users would actually want to share.

• Prompted GPT with metric combinations to generate archetype concepts
• Iterated on naming until each personality felt distinct and shareable
• Paired each personality with a spirit animal metaphor for visual direction

Designed a custom identity poster for each personality — layout and background in Adobe Illustrator, animal icons generated and refined in Midjourney, then composited together.

• Established a poster template system with personality-specific color palettes
• Generated animal icon concepts in Midjourney, selected and refined the best outputs
• Built each poster in Illustrator, ensuring each design felt cohesive as a set while being visually distinct

Built the web app to authenticate with Spotify, pull listening metrics via the API, run the personality classification, and display the result with the corresponding poster.

• Implemented Spotify OAuth and top-tracks/audio-features API calls
• Wrote classification logic mapping metric scores to personality types
• Built a shareable result card for social sharing

Getting Spotify to extend API access beyond 25 users required navigating their quota extension review process — the most unexpectedly complex part of the project.

• Iterated through multiple rounds of submission
• Approved after demonstrating clear user benefit and data handling practices