Скачать или смотреть Scaling User-Focused Foundation Models at Grab with Ray | Ray Summit 2025

At Ray Summit 2025, Nick Buehrer and Chongyu Zhou from Grab share how the company is building a powerful user embedding foundation model to understand the full, multi-modal journeys of users across Southeast Asia’s leading super app.

They explain how a single Grab user interacts across a wide range of services—ride-hailing, food delivery, payments, and more—creating a complex, interconnected behavioral footprint. To learn from this rich tapestry of signals, Grab set out to develop a unified model capable of capturing a holistic view of each user.

The result is a set of general-purpose user embeddings: numerical representations that serve as a universal feature set for numerous downstream applications. These embeddings eliminate the need for siloed, hand-engineered features and unlock more personalized, consistent, and scalable ML capabilities across the platform.

If you’re interested in multimodal modeling, foundation models for personalization, or large-scale production ML systems, this session offers a deep look at Grab’s approach.

Liked this video? Check out other Ray Summit breakout session recordings    • Ray Summit 2025 - Breakout Sessions  

Subscribe to our YouTube channel to stay up-to-date on the future of AI!    / anyscale  

🔗 Connect with us:
LinkedIn:   / joinanyscale  
X: https://x.com/anyscalecompute
Website: https://www.anyscale.com/