How Spotify & Snap Use ML for Recommendation Systems

Curtis leads the content moderation and understanding group at Snap and has 17+ years of industry experience in building both consumer and enterprise products. Most recently before Snap, Curtis led the core ML and data foundation group at Facebook/Meta, responsible for content understanding, measurement capabilities, and ML platform. Highlights of his work included spearheading and launching multiple 0→1 initiatives, such as on-device AI, graph learning, and AI governance, and he was the co-organizer for the FB NLP Summit for several years. Previously, Curtis was a founding member of Gemini at Yahoo!, revamping the ML ranking and retrieval stack for mobile sponsored search. Prior to Yahoo!, Curtis led ML efforts in the innovation center at SAP Labs, focusing on natural language understanding, in-memory & distributed computing, and representation learning.

Tony is the head of machine learning and vice president of engineering for personalization at Spotify. He leads a team of 100+ engineers and machine learners running Spotify's Home Page, Search Engine, Programming Platform, Voice Interface and User/Content Valuation. Tony also oversees the company-wide machine learning and AI strategy. Spotify’s ML strategy drives platform investments with new capabilities (across model training, serving, feature stores, logging and experimentation) as well as best practices in engineering and science. Previously, Tony was the director of machine learning at Netflix, where he launched improvements to many of its personalization algorithms. From 2001 to 2020, Tony was a tenured professor at Columbia University where he published hundreds of scientific articles with tens of thousands of citations. Tony holds a PhD from MIT.