Built inside the systems regulators scrutinize.
AI security architecture and governance, from someone who has owned it at cloud scale
I'm Mozhdeh, founder of Mozhie.ai. Before this I was an AI Security Architect on the Security CTO team at Oracle Cloud Infrastructure, where I owned the AI harness control framework for how AI tools and agents are used across engineering and led a Responsible AI control programme mapped to NIST AI RMF, ISO 42001, and the EU AI Act. Earlier roles at Salesforce, Sanofi Genzyme, and State Street taught me how regulated businesses actually adopt technology. I completed the Chicago Booth Chief AI Officer program. I'm based in Cambridge, MA.
How I work
Three habits from a decade in security architecture, applied to AI.
Controls before features
Every AI system gets a scoped identity, a permission boundary, and a kill switch before it gets a launch date.
Evidence, not assurances
Governance is only real if an auditor can trace it. I build the control mapping and the evidence trail together, so NIST AI RMF, ISO 42001, and EU AI Act alignment is provable, not claimed.
Executive language
Boards and CFOs need risk in dollars, timelines, and decisions, not model internals. I translate between the engineers building AI and the executives accountable for it.
See how the framework fits your programme
Agent identity at the core, pipeline security primitives as the shell, runtime monitoring on the surface, and AI safety as the foundation. Layered protection, built like a dome.
Let's talk
Telephone: +1 617 312 1252
E-mail: info@mozhie.ai
Address: Cambridge, MA