Neeraj Kumar Singh Beshane

Governing Generative AI in Production: Building Trustworthy and Secure AI Systems

Abstract:

Generative AI has moved from experimentation into production systems that support customer service, software engineering, healthcare, legal analysis, and enterprise decision-making. As adoption accelerates, governance maturity often lags behind the speed of deployment. This session examines how organizations can operationalize generative AI responsibly by treating AI behavior, not only software functionality, as something that must be governed, monitored, and secured.

The talk presents a practical governance and security framework for production AI systems. It covers AI-specific risks such as sensitive data exposure, hallucinated outputs, harmful bias, compliance failures, and attacks against prompts, inference pipelines, and AI infrastructure. It then maps these risks to concrete safeguards including risk-tolerance definition, clear accountability, AI lifecycle mapping, escalation procedures, identity and access controls, environment segmentation, encryption, input validation, output validation, and continuous monitoring.

The session also discusses operational excellence through MLOps and LLMOps, including prompt management, dataset and evaluation documentation, deployment automation, and cross-functional governance between engineering, security, legal, and business teams. It closes with compliance, privacy-by-design, ethical oversight, and human-in-the-loop review as core requirements for regulated and high-impact use cases. The key message is that strong AI governance enables innovation by making trustworthy deployment possible at scale.

Profile:

Neeraj Kumar Singh Beshane is a Staff Security Infrastructure Engineer at Parafin, Inc., where he works on Zero Trust security infrastructure, incident response readiness, business continuity, disaster recovery, and PII governance for a fintech platform serving major partners including Amazon, DoorDash, Walmart, and TikTok. He previously worked at Meta Reality Labs, Wayfair, JPMorgan Chase, and American Express, with 15+ years of experience across distributed systems, security infrastructure, DevSecOps, and AI-powered engineering tools.

His research and technical work focus on the security and governance of AI systems in production environments, including secure retrieval-augmented generation, enterprise GenAI governance, and runtime verification for autonomous AI systems. He has been recognized as a Gold Laureate in the Aureum Technology Awards (Season 3, NLP Achievement Award), Fellow of Hackathon Raptors, Fellow of IICSPA, IEEE member, and ORCID-registered researcher.

• GitHub: https://github.com/neerazz
• Google Scholar: https://scholar.google.com/citations?user=zfPkLR4AAAAJ