CFP Directory

Yuvraj Pradhan is a security researcher working on agentic AI, where he breaks the trust boundaries production agent stacks depend on. His work covers LLM red teaming, MCP security, and approval-provenance failures in the persistence and resume paths shared by agent frameworks, workflow engines, and CI/CD systems, along with research into running offensive security workloads on small local models instead of frontier infrastructure. He is first author of Springer Nature research on a 125M-parameter language model for STEM tasks, and has spoken at MCP Dev Summit Bengaluru and Mumbai, both Linux Foundation events, with the Bengaluru session featured in the Agentic AI Foundation's official recap.

I am an R&D Software Engineer at WSO2, where I build secure, scalable, and cloud-native enterprise software with a focus on API platforms, distributed systems, and AI-driven technologies. I hold a Bachelor's degree in Computer Engineering with a GPA of 3.95, and my work centers on the intersection of application security, enterprise AI, and modern software architecture.My research and engineering interests include Zero Trust security, AI agent security, API security, identity and access management, and the Model Context Protocol (MCP). I am passionate about helping organizations build trustworthy AI systems that securely interact with enterprise applications and cloud infrastructure.I have been selected as a speaker at FOST Munich 2026 (APIDays Munich), where I present on production-ready AI agents and the Model Context Protocol, and at FOST Toronto 2026, where I present "Zero Trust for AI Agents: Security Patterns Every MCP Server Needs," focusing on practical security patterns for securing AI agent ecosystems and enterprise MCP servers.Through my work and community contributions, I enjoy sharing practical knowledge, real-world engineering experiences, and emerging security practices that help developers and security professionals build resilient, secure, and production-ready software systems.

Alfonso De Gregorio is a globally recognised cybersecurity technologist, award-winning research artist, and strategic policy advisor. He has spoken at 25+ peer-reviewed int'l events across 5 continents, such as NATO Conference on Cyber Conflict, RSAC, and the leading hacker events. High-performance organisations engage him to spearhead relentless innovation across disciplines and fields, accelerate asymmetric advantage, and achieve peak confidence in today's interconnected operational environment—establishing Alfonso as a key figure shaping the discussion and practice of cybersecurity.Today he is Founding Director and Principal Investigator at Pwnshow, an interdisciplinary research agency investigating critical cybersecurity challenges at the complex technology-society nexus; CEO at Zeronomicon, a premium zero-day vulnerability acquisition platform; and, Member of the ETSI TC SAI (Securing AI), where he works towards ensuring the technical standards underpinning the EU AI Act are practical, effective, and innovation-friendly. At the forefront of the AI governance debate, his current work focuses on the dual-use dilemma of open-weight AI and how the proliferation of powerful models impacts the cyber threat landscape. Active in the legislative and standardization trenches as much as at the terminal prompt, he provided expert technical consultation to the European Commission regarding the EU AI Act. He successfully advocated for the "substantial modification" clauses in the GPAI Code of Practice, protecting open model developers from undue liability.

My name is Đoàn Đình Toàn, but you can call me Twan. I'm a Threat Intelligence Analyst at VCI-CTIP (Vietnam Cybersecurity Intelligence), where I research and publish vulnerability advisories and threat intelligence reports. My current research focuses on code-generation injection vulnerabilities across the JavaScript ecosystem.