data: (plural, datum) — Sanskrit, Latin. To give. To impart.
“Vision : enable organisations to build AI worth trusting, govern what it does, and grow the people who run it.”
Three things, and they hold together: data science, agent governance, and corporate training. Design the system, keep it inside its remit, and leave behind a team who can run it without you.
My work is architecture, review and enablement: deciding what should be built and why, examining what has already been built, and bringing a client’s own people up to the standard needed to own it. I have built these systems myself, which is the reason the advice is worth anything.
I am Director at LeewayHertz and Enterprise Generative AI Solution Architect at its parent, The Hackett Group. Day to day that is enterprise generative AI work for organisations where a mistake is a regulatory event rather than an embarrassing one.
Data science
Twenty years in emerging technology, the last several in machine learning and generative AI. Framing the question, choosing the method, designing the evaluation, and knowing when the honest answer is that the data will not support the decision being asked of it.
Agent governance
This is my newest focus, and I will be straight about where it stands. The tooling and control specifications the industry is converging on are still being written — much of it lands through 2026. I am building this practice alongside those standards.
What I bring is the reason I went looking. I built prompt-level guardrails and agent evaluation in production, and I saw where they fail: a prompt guardrail makes an unwanted action unlikely, not impossible. Governance moves the decision in front of the tool call — agent identity and delegation chains, privilege tiers, approval gates, kill switches, and an audit trail someone else can verify.
In practice that means reviewing an agentic implementation you already have and telling you plainly where the gaps are: which agent can reach further than anyone intended, what happens when one is wrong, whether you could reconstruct a decision six months later, and how the controls line up against the regime that governs your sector. Across corporate, enterprise, data and application-level governance, the aim is the same — an agent’s blast radius should be a decision you made, not a discovery you make later.
Corporate training
I rebuilt the data science vertical at Northcap University: over 2,000 hours of courseware, the entire faculty upskilled including PhDs, and an undergraduate programme rated fifth pan-India two years running with a 99% placement rate. Before that I co-founded India’s first game programming and AI school.
When I am not working, I am vibe coding, researching emerging technologies, studying probability and analysing risk markets. I am an Indian national who speaks Hindi, English and Coding.
