Partnerships

Request your invite

Call to action
Your text goes here. Insert your content, thoughts, or information in this space.
Button

Back to speakers

Sunondo
Ghosh
Chief Technology Officer
iink Payments
Sunondo Ghosh, Ph.D., is the Chief Technology Officer at iink Payments, where he leads engineering, AI, and product innovation to transform insurance claim payments through intelligent automation. With over 25 years of experience, Dr. Ghosh has built and scaled technology organizations at companies ranging from startups to industry leaders like Intuit, 2Q.ai, and Percipient.ai, specializing in applying AI to drive business growth and operational efficiency. An expert in Generative AI, prompt engineering, model fine-tuning, and agent-driven systems, Dr. Ghosh has developed AI applications for fields including insurance, financial services, and national security. His work spans creating production-grade AI agents, building large-scale analytics platforms, and designing intelligent workflows that improve decision-making and speed. He is passionate about bridging cutting-edge AI with real-world impact, particularly in complex, regulated industries. Dr. Ghosh holds a Ph.D. and M.S. in Computer Science from the University of Pittsburgh and a B.Tech. in Computer Science and Engineering from IIT-BHU, India. Throughout his career, he has led high-performing teams across software development, security, data engineering, and DevOps, and continues to champion innovation at the intersection of AI, fintech, and insurtech.
Button
14 April 2026 11:00 - 11:30
AI agents gone rogue! And how to rein them in
As AI agents move from demos to production, some have already gone off the rails - triggering outages, burning money, or taking actions no one explicitly approved. This session examines real incidents where agents failed in surprising ways, and the technical and organizational factors behind those failures. We’ll dig into how misaligned objectives, insufficient constraints, and poor observability lead to “rogue” behavior. The goal is to leave with concrete techniques for designing, testing, and deploying agents that stay aligned under real-world conditions.