I lead enterprise AI and data product portfolios in global financial services, after twenty-two years across banking, aviation, and consulting. My work is turning AI ambition into value a business can measure — which only happens when people trust what has been built.
A small number of principles, drawn from twenty-two years in regulated industries. They apply as readily to an airline as to a bank, and they decide far more value than any choice of technology does.
Responsible innovation compounds.
Governance and evaluation designed in at the outset cost a fraction of what they cost later, and they let teams move faster because the guardrails are already understood. Trust accumulates. Rework does not.
Technology should extend people, not replace them.
The most valuable systems I have led made expert judgment worth more, not less. Automation earns its place when it removes what is repetitive and returns skilled attention to the decisions that actually move the business.
Value appears at adoption, not at launch.
A capability the organization does not depend on has created nothing, whatever the announcement said. I hold my teams to that standard, and I hold myself to it first.
The framework
The Adoption Gap
Every enterprise AI program passes through four gates. Most organizations clear the first two comfortably. Value is created only after the fourth. The distance in between is where transformation is won or lost.
01
Mandate
Leadership commits and funds the work.
Ambition is rarely the constraint.
Where most programs stop
02
Pilot
The capability is proven in controlled conditions.
Encouraging, reversible, and often where progress quietly halts.
03
Production
It meets the standards of risk, control, and integration.
Where governance and evaluation prove their worth.
04
Adoption
The organization depends on it in daily operation.
The only stage that appears in results.
That distance is where the value is created or lost, and closing it is the work I have spent my career on. It comes down to sequencing, accountability, and the willingness to decline good ideas at the wrong moment.
In a world racing toward automation, the greatest innovation will be remembering what makes us human.
Swati Deepak Kumar · Forbes Finance Council, 2025
What I do
Four ways value gets created.
Alongside my work at Citi, I take on a small number of engagements each year, chosen for the seriousness of the question rather than the size of the organization.
Advisory
Clarity on where AI investment will return, and a sequence for getting there. For boards setting direction, and for programs stalled between pilot and production.
A ranked portfolio, with the investment case for each bet
Governance and evaluation that satisfy risk without stalling delivery
An independent read on whether a program will reach production
Building
A working capability in production, owned by your team. For organizations that need something that runs, not a recommendation and not another prototype.
Retrieval, agentic, and conversational systems built for regulated environments
Evaluation, observability, and controls designed in from the start
Delivery through to adoption, with ownership handed over cleanly
Training
Teams that can make and defend AI decisions without outside help. Capability that stays in the organization after the engagement ends.
Executives who can tell a real AI investment from a costly one
Product and data teams fluent in discovery, evaluation, and shipping
Governance and responsible deployment practice teams can run themselves
Speaking
An audience that leaves with something usable on Monday. Keynotes and panels on enterprise AI, governance, and the future of financial services.
Responsible innovation as competitive advantage
Governing AI without slowing it down
Keynotes, panels, podcasts, and leadership forums
Writing
From my pen.
On responsible innovation, AI governance, and what the next wave of technology means for regulated industries. Published with the Forbes Technology and Finance Councils, and at greater length on LinkedIn.
I have spent twenty-two years in enterprise technology, across financial services, banking, aviation, and consulting. What has stayed constant through all of it is a belief that trust, not capability, is the scarce resource — and that it is earned slowly, in how a system behaves when it is inconvenient.
Very few organizations struggle with ambition. They struggle with sequencing, with accountability, and with the discipline to decline good ideas at the wrong moment. In my experience the difference between a program that changes an institution and one that quietly stalls is almost never the technology. It is whether leadership was honest about what the organization could absorb, and patient enough to see it through.
Responsible innovation is not a constraint. It is a competitive advantage.
That belief shapes how I lead. I would rather fund four things properly than sponsor twenty and defend the portfolio in a steering committee. I expect governance to be designed at the start rather than retrofitted under pressure. And I would rather deliver one capability an organization comes to depend on than a shelf of pilots nobody misses.
I write, teach, and speak because I believe the next decade of artificial intelligence will be decided less by capability than by governance — by whether institutions can be ambitious and responsible at the same time. That is where the value will be created, and where it will be squandered.
The path here
2001 – 2005
Engineering
Writing and shipping software. A system is judged in production, not in review — and the last ten percent decides whether anyone comes to rely on it.
2005 – 2012
Leading teams
Running engineering and product teams across aviation and telecom. Adoption is a discipline of its own, and no rollout survives without named owners.
2012 – 2019
Consulting and founding
Large-scale implementations, then four years running my own venture. Changing how thousands of people work costs far more than building the thing they will use.
2020 – present
Executive leadership
Enterprise integration, then firm-wide AI and machine learning governance, now an enterprise AI and data portfolio. At this scale the hardest problems are never the technical ones.
Credentials
MIT — AI Strategy and Leadership, 2026
MIT — Designing and Building AI Products and Services, 2025
Pragmatic Institute — Product Manager, Product Master, and Certified Product Marketer, 2022
Project Management Professional (PMP) — PMI, 2021
Certified Scrum Product Owner and ScrumMaster — Scrum Alliance, 2019
PRINCE2 Foundation — APMG International, 2012
Airline Revenue Management — IATA, 2010
Master of Computer Applications — Rajiv Gandhi Technical University, 2002
Diploma in Advanced Computing — C-DAC, 2003
Contact
Get in touch.
For advisory engagements, training programs, speaking invitations, and writing collaborations. A few sentences on what you are working on is enough to start.
