• Launched multi-objective, multi-stakeholder online recommendations in programmed music
playlists (e.g. Spotify radio). The recommender system is used by Spotify's ~600 million monthly
active listeners, with a 200ms P99 latency.
• Enabled the highly profitable Discovery Mode promotion program within recommendations,
tripling the available capacity for the program, and improving core metrics.
• Developed and successfully executed a technical strategy for the 150-person business unit to
successfully adopt ML across the technical stack and engineering practices.
• Shipped 5-10% user satisfaction metric improvements across Spotify's playlists.
• Oversaw the development and deployment of ML-driven recommendations within the AI DJ.
• Managed a 22-person ML-focused organization, including multiple engineering managers
and staff engineers.
• Grew and mentored multiple technical leaders across Spotify