Mindful ML at Headspace Health: How to Practice Personalization Without Sacrificing Your Values
Headspace Health is a values-first company: through 10 years of innovating the use of technology in meditation and mindfulness, the focus has never shifted from making sure that our approach to technological implementation is as mindful as our content. Over the last year, much of that innovation has centered around ML-powered personalization: from in-app content recommendation to push notifications to AI-driven dynamic welcome flows. But implementing ML/AI solutions brings unique challenges: personalization is inherently intimate and demands great sensitivity in the ways we use and communicate member data. Additionally, de-biasing datasets and models is a huge priority, since both domains are prone to many types of bias. In this talk, we’ll explore how Headspace Health balances the intimate and data-centric process of building ML solutions while maintaining our brand identity as a mindful actor.
Matt Linder is a Senior Machine Learning Engineer at Headspace Health, where he develops and implements models, maintains internal tools and libraries, and builds out ML-powered prototypes. He loves collaborating with various squads, teams, and product owners to develop AI-powered solutions. In another life, Matt was and is a touring professional classical guitarist with his group Mobius Trio. They focus on developing new music for plucked strings - collaborating with composers, other performers, and institutions to do so.