Abstract
Autonomous agents operate in a continuous loop: perceive context → reason → execute tools → observe. Each edge creates distinct attack surfaces. This talk maps vulnerabilities—memory poisoning in context ingestion, goal hijacking during reasoning, blind execution at the action stage. You'll learn defensive patterns for each edge: provenance gates, dual-model critics, risk-weighted human oversight, ephemeral credentials, and sandboxed execution. Illustrated with real industry incidents.
Speaker
Sriram Madapusi Vasudevan
Senior Software Engineer @AWS Agentic AI, Previously Core Team @AWS SAM, AWS Cloudwatch, Core Developer @Openstack
Sriram Madapusi Vasudevan is a Senior Software Engineer at AWS focused on building AI agent-ready developer experiences. Over the past decade, he has worked on large-scale platforms such as AWS CloudWatch, Rackspace Cloud Queues/CDN and open-sourced developer tooling such as AWS SAM CLI, AWS Lambda Builders, and created the AWS Homebrew tap.