What well-being sounds like, in 600 hours of interviews

Machine Learning Engineer · Sep 2024 – Feb 2025

F1, BERT

0.89

F1, BERT

of interviews transcribed

600 h

of interviews transcribed

place, Well-Being track

1st

place, Well-Being track

Supervised by a Full Professor of Genetics and Wellbeing on XplorIT's board, my team of four turned ~600 hours of interviews, mostly with sailors, into a labelled corpus and compared three model families on it.

At a glance

• Scraping and transcription pipeline: BeautifulSoup, Selenium, OpenAI Whisper.
• LIME and SHAP to show which expressions the models tied to well-being, the deliverable our supervisor needed.
• Hosted on Hugging Face with an interactive demo. 1st place in the Well-Being track.