Architecting for agentic AI development on AWS

A serverless, event-driven architecture using computer vision and generative AI is developed to automate safety monitoring across hundreds of facilities. The system captures near real-time visual data from fixed camera networks, detects potential safety hazards, and scales across thousands of cameras, while protecting personal identifiable information (PII) by blurring faces and identifiable features. The architecture is designed to learn and improve detection accuracy continuously, using a hierarchical role-based access control structure and synthetic data generation with generative AI tools like GLIGEN to reduce site onboarding time and data labeling requirements.

To implement powerful disaster recovery capabilities on AWS, users can leverage native AWS services like AWS Backup and AWS Elastic Disaster Recovery (AWS DRS), in conjunction with AWS Resilience Competency Partner solutions like Arpio. These tools enable cross-Region and cross-account backup and recovery, streamlining the process and reducing manual effort. By centralizing and streamlining backup and recovery processes, users can enforce consistent protection across resources and achieve a high level of resilience. Key to achieving a comprehensive disaster recovery solution is protecting not just data, but also compute resources, networking, and other infrastructure components. Users can achieve near-zero data loss and recovery in minutes or less using AWS DRS, and can also recover entire workloads, including complex compute services and networking, using tools like Arpio.