I77537 StackDocsTechnology
Related
How to Conquer the Revamped Endgame in Path of Exile 2’s Return of the Ancients UpdateHow to Set Up and Use Orion for Linux v0.3 Beta with Content Blocker and Download ManagerEri Takes Center Stage: New My Hero Academia Anime Spin-Off Short AnnouncedSafari 27 to Introduce AI-Driven Automatic Tab Grouping at WWDCShould You Build an AI Chatbot of Your Ex? The Surprising TruthModeling Complex Systems: How HASH Brings Simulations to EveryonePython 3.14.3: Key Updates and New Features ExplainedMastering Software Project Management: Lessons from The Mythical Man-Month

Meta Unveils Friend Bubbles: The Engineering Behind Social Discovery at Scale

Last updated: 2026-05-18 19:56:15 · Technology

Meta has launched a new feature called Friend Bubbles for Facebook Reels, enabling users to see which videos their friends have watched and reacted to—a seemingly simple addition that required monumental engineering efforts to scale to billions of users.

The feature, which highlights Reels your friends have engaged with, is now rolling out globally. Engineers Subasree and Joseph, speaking on the Meta Tech Podcast, revealed the complex machine learning model and cross-platform challenges that made the feature possible.

Background

Friend Bubbles began as an experimental project to boost social discovery on Reels. The engineering team faced scalability hurdles as the model needed to process interactions from over a billion users in real time.

Meta Unveils Friend Bubbles: The Engineering Behind Social Discovery at Scale
Source: engineering.fb.com

“The path from a prototype to a feature serving billions is never straightforward,” said Subasree, a software engineer on the Facebook Reels team. “We had to redesign our machine learning pipeline multiple times to handle the volume.” The team also discovered unexpected behavioral differences between iOS and Android users, affecting how friend bubbles were populated.

“On iOS, users tend to interact more with Reels in private groups, while Android users favor public content,” explained Joseph. “That forced us to build separate ranking models for each platform.”

What This Means

The launch of Friend Bubbles signals Meta’s push to integrate social signals into content discovery. By showing what friends are watching, the feature aims to increase time spent on the platform and strengthen peer-to-peer engagement.

Meta Unveils Friend Bubbles: The Engineering Behind Social Discovery at Scale
Source: engineering.fb.com

“This is a major shift from algorithmic recommendations alone,” said an industry analyst. “It turns Reels into a more communal experience, which could drive higher retention.” For engineers, the project highlights the complexity of building features that feel simple but require deep infrastructure work—particularly in machine learning and cross-platform consistency.


You can listen to the full Meta Tech Podcast episode on Spotify, Apple Podcasts, or Pocket Casts. Send feedback to Meta on Instagram, Threads, or X. Explore career opportunities at Meta Careers.