Serving 2 Million Models Without Melting: Scaling the Hugging Face Hub
Hugging Face hosts over 2 million public models, 500,000+ datasets, and serves 13 million users across 50,000+ organizations, including over 30% of the Fortune 500. That growth didn't come with a manual.In this talk, we'll pull back the curtain on the infrastructure decisions that kept the Hub fast and reliable as traffic grew by orders of magnitude. We'll dive into why we chose MongoDB Atlas as our core data layer, how its document model maps naturally to the messy reality of ML model metadata, and what it took to keep p99 latency low when every request hits a catalog of millions. We'll also cover the trade-offs we faced, the things that broke along the way, and what "lean operations" actually means when your platform serves a third of the Fortune 500. Expect real architecture decisions, real numbers, and lessons you can take back to your own stack.
About the AI Architects: Show my Workflow Track
AI Architects: Show my Workflow sessions at AI Engineer World's Fair 2026 in San Francisco.
When
Tuesday, June 30, 2026
1:30 PM - 1:50 PM·20m
Where
Leadership 2 · Room 3020
Capacity: 550 attendees
Sign in to add this talk to your schedule.
Speaker
Arek Borucki
Hugging Face
@_Aras_B
Arek Borucki is a Machine Learning Platform & Database Engineer at Hugging Face, where he helps keep the infrastructure behind one of the world's largest open-source AI platforms running at scale. He is the author of MongoDB in Action 8.0 and co-author of Mastering MongoDB 7.0. With over 10 years of experience in SRE, Kubernetes, AWS, GCP, and managing MongoDB in production, from 100TB+ sharded clusters to cloud-native deployments, he brings deep expertise in databases, platform engineering, and infrastructure at scale.