Meetup 10 : An Evaluation Harness with Amazon Bedrock AgentCore

About this event

As AI agents become more capable and increasingly move into production, evaluating their behaviour consistently becomes an important part of building reliable applications. In this session, Manoj Kumar Sashikumar, Associate Director at DTCC, will take us through a practical approach to evaluating AI agents using Amazon Bedrock AgentCore Evaluations. We’ll look at how evaluation can be used to assess agent behaviour, measure quality and performance, validate changes, and continuously monitor deployed agents for regressions. AWS currently supports both targeted/on-demand evaluation and continuous online evaluation, along with built-in and custom evaluators. What you’ll learn • Why AI agent evaluation is different from traditional application testing • How to evaluate quality, safety, and performance • Building an evaluation harness with Amazon Bedrock AgentCore • Evaluating agent traces and interactions • Using evaluations for regression testing • Understanding continuous evaluation after deployment • Practical demonstration of an agent evaluation workflow 🎤 Speaker Manoj Kumar Sashikumar Associate Director 📅 Event Details AWS User Group Puducherry – Meetup #10 📅 October 4, 2026 ⏰ 4:00 PM – 5:00 PM IST 💻 Virtual (Gmeet) Who should attend? This session is useful for developers, data engineers, AI/ML practitioners, cloud engineers, architects, students, and anyone exploring AI agents and production-ready GenAI applications. Come with your questions and learn together with the community. 🚀