I got to enterprise AI architecture by way of security, not the other way around — and that order turns out to matter.
I started as a software engineer building machine learning, computer vision, and ETL systems (Reezanorp Technologies, 2016–2022). From there I moved into security — leading Zero Trust segmentation and a SASE migration covering 5,000+ users as Technical Security Lead at Arcstream Technologies (2022–2025).
Then generative AI collided with enterprise security, and I ended up at the center of it. At Levo.ai, working on a national-scale insurance data ecosystem — one that aggregates sensitive records from across the country’s insurance industry — I architected the AI Trust Fabric: a Zero Trust framework (NIST 800-207, structured as PAP/PDP/PEP/PIP) extended with an Adaptive Trust Engine and AI governance controls, purpose-built for a world where the thing requesting access to your data might be an autonomous agent, not a person. That work is now a registered design with the Government of India.
Separately, I’m a co-author on MCP Guardian, a peer-reviewed paper proposing a security layer for MCP-based AI systems — independent research with collaborators across eight companies, and the closest thing to a public statement of how I think agent tool-calls should be governed.
I’m currently AI Engineer Advisor at NTT DATA, where I lead architecture and technical direction for a team of senior engineers, AI engineers, and data scientists across 8–10 enterprise AI modernization programs. Most of that work runs through Syntphony, NTT DATA’s productized technology ecosystem — I architect AI modernization across Syntphony AI, Syntphony Open Telecom Networks, Syntphony Insurance Cloud Migration, Syntphony Stations, and Syntphony Commerce & Payments.
In practice that means conversational knowledge systems, agentic platforms built on MCP and A2A, legacy mainframe modernization, and cross-domain operational intelligence. My role is architecture and technical leadership: defining the systems, making the trade-offs, and setting the direction the team builds toward.
The throughline across all of it — security architecture, Zero Trust for AI agents, and now agentic AI platforms at enterprise scale — is the same question asked in different contexts: how do you let a system act autonomously without giving up control over what it can touch? That’s what I mean by Zero Trust AI, and it’s what most of my writing and work here is about.
How I lead
My job is architecture and technical direction, not implementation. The team — senior engineers, AI engineers and data scientists — builds; I’m accountable for the shape of what gets built and for the decisions that are expensive to reverse.
In practice that means a few things. Architecture reviews and decision records, so that the reasoning behind a choice survives longer than the meeting it was made in — the trade-off is usually the interesting part, not the conclusion. Reusable platform work over per-engagement rebuilds, which is the entire premise of the agentic platform: when five teams are solving tool integration, identity and audit separately, the fix is a platform layer, not five better implementations. And stakeholder alignment, because most architecture decisions at this scale are really negotiations about constraints someone else owns.
The part I care most about is making the reasoning legible. An architecture nobody on the team can explain back to you isn’t a shared architecture — it’s a bottleneck with your name on it.
Beyond listening
In 2020 I started a YouTube series called Beyond Listening, taking apart Indian film music one song at a time — an Ilaiyaraaja love song that turns out to be in 7/8, the bossa nova harmony hiding under a road-trip number.
The name has quietly become the most accurate description of how I work. Listening is the passive default: you let something wash over you and register that it worked. Going beyond it means asking what actually produced the effect — which structure, which decision, which trade-off someone made on purpose.
That’s the same question I ask of a system. Why this boundary and not that one. What was given up here, and in exchange for what. An architecture and a song are both things that appear to simply work until you look at the mechanism.
So the guitar and the travelling aren’t a change of subject from the architecture. They’re the same habit with the surface swapped out.