Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about Meta's PAPAYA Federated Analytics Stack in this 17-minute conference presentation from NSDI '25. Discover how this system addresses the key limitations of existing federated analytics approaches by combining privacy, scalability, and practicality in a comprehensive solution. Explore the technical architecture that leverages trusted execution environments (TEEs) and optimizes on-device computing resources to enable federated data processing across large device fleets while maintaining robust privacy safeguards. Understand the distinction between federated analytics for statistics and monitoring versus federated learning for machine learning workloads. Examine how PAPAYA overcomes traditional challenges including compromised accuracy, limited flexibility for data analytics, and scalability issues in cross-device federated analytics systems. Gain insights into the engineering decisions and privacy measures that ensure minimal data transmission off-device while achieving high standards of data protection in distributed computation paradigms.
