Detecting AI-Driven Automation in E-Commerce Traffic Using Real-Time Behavioral Signals
Friday, April 17, 2026 · 10:55 AM – 11:25 AM · Auditorium 3
Recent advances in large language models and browser-based automation frameworks have enabled a new class of adaptive systems capable of interacting with digital platforms in ways that closely resemble human users. In large-scale e-commerce environments, these AI-driven agents are increasingly used for automated purchasing, credential stuffing, inventory scraping, and API abuse, often bypassing traditional bot mitigation systems designed to detect scripted or non-interactive traffic.
This talk presents a production-oriented case study on applying machine learning–based behavioral signal analysis to identify synthetic user activity in high-volume platform traffic. It will discuss the design of scalable detection pipelines that analyze session-level interaction patterns, distinguish between legitimate user sessions and automation-mediated workflows, and support real-time mitigation strategies without degrading user experience.
We will examine challenges encountered when deploying these systems in live environments, including evolving automation signatures, adversarial evasion techniques, latency constraints, and privacy-aware signal interpretation. Attendees will gain practical insight into how behavioral ML systems can be operationalized to improve platform integrity, analytics reliability, and fair user access in modern digital marketplaces.
About the Speaker
Shashwat Jain
Amazon, Sr. Software Development Engineer
Shashwat Jain is a Senior Software Engineer at Amazon specializing in the development of machine learning–driven behavioral detection systems for large-scale digital platforms. His work focuses on identifying automated and synthetic user activity in environments where traditional scripted bots increasingly coexist with adaptive AI-driven agents operating through browser-based workflows. He has experience designing scalable ML pipelines and rule-based detection frameworks that analyze session-level interaction patterns, detect automation signatures, and enable real-time mitigation strategies in high-volume traffic ecosystems. His professional interests include behavioral signal intelligence, privacy-aware ML deployment, and the infrastructure challenges associated with maintaining platform integrity in environments with mixed human and agent-mediated participation.