What if you could build an intelligent, searchable photo gallery, without managing a single server? In this talk, we'll take you behind the scenes of how we built an AI-powered image gallery using a 100% serverless and event-driven architecture on AWS.

Every time a user uploads a photo, a series of lightweight, asynchronous workflows is triggered automatically. Images are compressed on the fly for optimal performance and then indexed using Amazon Rekognition to detect and store facial data. This enables users to search through their gallery using just a face. The best part - no manual tagging or sorting required.

We’ll explore how services like S3, Lambda, Rekognition, and Cognito come together to build a scalable, low-maintenance system that’s both cost-effective and production-ready.

Beyond the technical architecture, I will also share performance wins, cost breakdowns, and lessons learned