Examining emerging attack surfaces created by agentic AI systems and the engineering controls required to deploy them safely at scale.
Speaker
Tejas Pravinbhai Patel
Agenda
6:00pm - Introduction
6:10pm - Speaker
7:10pm - Questions and answers
7:30pm - Close
Synopsis
As AI systems evolve from passive models into autonomous agents capable of invoking tools, retaining state, accessing data, communicating with external services, and executing actions across distributed infrastructure, they introduce an entirely new class of security risks.
This session will examine the emerging attack surface created by production agentic AI systems and the engineering controls required to deploy them safely at scale.
The session will explore practical security challenges including delegated identity, least-privilege tool access, prompt and data injection, memory poisoning, secrets management, sandboxing, approval boundaries, auditability, anomaly detection, runtime monitoring, incident response, and containment of unintended autonomous actions.
Drawing on lessons from large-scale production systems, the discussion will focus on moving beyond model-level security toward protecting the complete distributed environment in which AI agents operate.
Attendees will leave with a practical framework for thinking about identity, authorisation, observability, trust boundaries, and failure containment when designing production-grade agentic AI systems.
About the speaker
Tejas Pravinbhai Patel is a Senior Software Development Engineer at Amazon working across artificial intelligence, large-scale distributed systems, e-commerce technologies, trustworthy AI, data platforms, and production-grade intelligent systems.
He is Chair of an ACM Chapter and Chair of the IEEE Computational Intelligence Society Dallas Chapter. He also serves as an Industry Adviser to the BCS Computer Science Academics Specialist Group and participates in the IEEE P3982 Working Group developing the IEEE Standard for Vision-Language Models Evaluation.
Tejas has authored more than 40 peer-reviewed research publications and completed more than 250 peer reviews for international journals and conferences. He also serves as a judge for the CES Innovation Awards 2027 and has contributed as an international speaker, keynote speaker, session chair, technical reviewer, and technology judge across professional and research communities.
His technical interests include agentic AI, trustworthy AI, distributed systems, AI infrastructure, software reliability, cloud architectures, and the security and operational challenges involved in deploying autonomous intelligent systems in production environments.
Our events are for adults aged 16 years and over.
This meeting is conducted in accordance with the BCS Code of Conduct for Meetings.
BCS is a membership organisation. If you enjoy this event, please consider joining BCS. You’ll be very welcome. You’ll receive access to many exclusive career development tools, an introduction to a thriving professional community and also help us Make IT Good For Society. Join BCS today
Please note, if you have any accessibility needs, please let us know via groups@bcs.uk, and we’ll work with you to make suitable arrangements.
BCS privacy notice: your data will be processed by BCS in accordance with our data privacy notice.
Photography: by attending this event, you may be photographed or filmed. Please speak to a member of staff if you do not wish to be included.
This event is brought to you by: Information Security specialist group
