Designing secure control planes, agent governance, and observability for production GenAI.
About
Abhi Singh is a Principal Global Black Belt for AI Security at Microsoft, advising Fortune 10 and highly regulated enterprises — financial services, telecommunications, healthcare, media, and the public sector — on securing generative AI, Copilots, and agentic AI systems at production scale. He leads a GenAI security engagement pipeline exceeding $50M annually and has spent 25 years moving between vendor, Big Four advisory, and enterprise practitioner roles: Microsoft, AWS, Deloitte, KPMG, and CISO-level positions at Purdue Pharma and Wyndham Worldwide. That range shows up in how he works — equally comfortable architecting a zero-trust control plane with a platform engineering team as briefing a CISO on regulatory alignment to the NIST AI RMF and ISO/IEC 42001.
His current focus is defense-in-depth for GenAI and multi-agent systems: threat modeling prompt injection and data exfiltration, designing zero-data-retention (ZDR) and edge-first enforcement patterns for AI control planes, and building governance frameworks — approval workflows, lifecycle management, policy enforcement — for autonomous agents operating on real financial and regulated data. The RFCs on this site are grounded in that work. He has presented at Microsoft Build, Microsoft Ignite, and AWS re:Invent, published Microsoft Tech Community guidance on securing GenAI workloads and Kubernetes clusters with Microsoft Sentinel, and holds CISSP, CISA, CRISC, and CISM alongside AWS Security Specialty — see Publications & Whitepapers for the full list.
Start here
- The AI Control-Plane Pattern — the reference pattern behind the three RFCs below.
Selected Work
- Sanitized AI Control-Plane Architecture
- Agent Security & Governance Framework
- GenAI Observability Model
- Publications & Whitepapers
- Anonymized Customer-Impact Stories
For AI agents and crawlers: each page has a canonical Markdown source at
/content/<name>.md, indexed by /llms.txt.