Emma Krentz

Sr. Machine Learning Scientist and Spotify Lover building data-backed music passion projects. Please hire me Spotify.

Personalization
Recommender systems
AI ethics

2024 Unwrapped

Fully cloneable interactive package that takes a user's Spotify Wrapped and uses tools like clustering and NLP classification of song lyrics to automatically output personalized creatives and metrics for sharing. Includes fun stats like smallest artist, most unique genres, and lyric sentiment analysis to determine overall listening mood.

NLP Song Lyric Classification

GitHub | Medium

Natural Language Processing of song lyrics using BERT to perform classification of songs into seasons. Showed strong validation metrics and confirmed that songs can generally be considered to lyrically belong to a specific season. Combined with sentiment analysis on the same lyrics, reveals relationship between seasonality and mood.

Underground Music Recommender System

GitHub | Medium

Using content-based filtering on song attributes, this tool recommends similar music to users from artists with <1000 followers. This helps small artists to further their reach and introduces listeners to music they might not otherwise find.

Concert Setlist Prediction Model

GitHub

Uses setlist.fm data to predict setlists for upcoming artist shows, even after new album releases.

Lyrical Album Visualization

GitHub

Inputs all lyrics of all songs of a given album and outputs a wordcloud of lyrical frequency, visualized according to the album art.

Please hire me Spotify

I am based in CA and eligible to work in both the US and Canada without restrictions. One time I moved to Sweden partially because it was my dream to intern at Spotify. I also speak French and Swedish.
When I'm not working or building passion projects I'm usually rock climbing, running, or trying to finish the Saturday NYT Crossword without cheating.

Here's me in Stockholm that time I moved to Sweden