Rootconf 2025 Annual Conference CfP

As large language models (LLMs) increasingly integrate into real-world applications—from querying internal databases to enabling autonomous agents that conduct online transactions, their potential seems limitless. In this session, we delve into the critical challenges organizations face when embedding LLMs into modern systems, such as prompt injection attacks, accidental API key exposure, and unauthorized use of sensitive customer data for model training.

We will also explore the often-overlooked fact that traditional security flaws within broader system architectures—such as insecure APIs or poor access control, can influence LLM behavior, potentially leading to harmful or unexpected outputs.

Attendees will leave with a practical understanding of the key vulnerabilities to watch for, guidance on securing sensitive data during LLM interactions, and a roadmap of best practices for designing AI-enabled applications that are not only innovative but also trustworthy and resilient. Whether you’re building chatbots, customer-facing agents, or internal AI tools, this session offers essential insights for ensuring your LLM-powered applications meet the highest standards of security and data governance.

Developers and Security Researchers

Vandana Verma Sehgal is a distinguished cybersecurity leader, currently serving as the Security Relations Leader at Snyk. With over 19 years of experience, she has expertise spanning application security, DevSecOps, cloud security, and product security. She is also a member of the OWASP Global Board of Directors and has held the Chair position.

Beyond her professional roles, Vandana is a passionate advocate for diversity and inclusion in cybersecurity. She founded InfosecGirls and InfosecKids, initiatives aimed at empowering women and children in the field. Her commitment to community building is evident through her involvement in organizing conferences like BSides Delhi and her participation in various diversity-focused communities.