AI InitiativeAI InitiativeThe architect
Research · Architecture · Building
From data and decisions
to intelligence that grows.
My work has moved through data, research, products and systems for intelligence working together. AI Initiative is where I pursue what comes next.
The evolution of humans and AI
AI is changing what it can be for people. The question I care about is how that evolution can stay a partnership, with each side making the other more capable.
AI waited for instructions and did one task.
AI responds, drafts and suggests, then forgets.
AI that remembers, learns from outcomes and grows with the people it works with.
This is co-evolution: human and machine capability rising together. I measure success by what people can now do.
Explore the full vision ↗Experience that connects
From client work and visualising risk to R&D, product leadership and AI architecture. Some roles ran alongside one another; the timeline keeps those connections visible.
AI Initiative began in January 2024 and the Fifth9 role in May 2024; the Insight Delivered CTO role continued until June 2024. Client BI projects for Context BI, RSA and Canopius were part of the Insight Delivered roles.
2001–2013 · Data and decisions
Early work spans TCS client work for NSE, client and contract work for Morgan Stanley, BI and ETL roles, and advanced visualisation at Credit Suisse.
These are the data and decision-making foundations of the journey: integrating information, making complex subjects understandable and supporting people working with it.
Data integration · Business intelligence · Risk visualisation2013–2024 · Research and product leadership
I co-founded Insight Delivered and progressed from Director R&D to CTO. The work included technical direction, global teams and the development of AI and machine-learning products such as AutoML, DeX and Strategic Finance.
Client projects within those roles included Qlik work through Context BI and BI strategy, architecture and Centre of Excellence programmes for RSA and Canopius, connecting product work with client settings.
R&D · Product development · Technical leadershipTCS
Client work for the National Stock Exchange.
Morgan Stanley
Financial-services client and contract work.
Project Brokers · Zensar Technologies · Unipro
Business intelligence and data integration roles.
Credit Suisse
Basel III risk visualisation.
Insight Delivered
Co-founder. Director R&D: Apr 2013–Sep 2021. CTO: Sep 2021–Jun 2024. Technical direction, global teams and AI/ML product development, including AutoML, DeX and Strategic Finance. Client projects: Qlik work through Context BI for Nationwide/Kofax, Morgan Stanley and HSBC (2013–2017); BI strategy, architecture and Centre of Excellence for RSA (2017–2019) and Canopius Group (2020–2023).
AI Initiative Ltd
Independent learning research, poker and Matiz, games and writing.
Fifth9
Based at HSBC, in Developer Experience. I architect Conduir, an agentic, multi-agent AI architecture in which people and AI agents work together. It composes models and tools for each task, keeps memory that outlasts any single model (Cortex), maps how complex systems fit together (GraFOSa), explores ideas through many perspectives (Manthan), and turns each reviewed outcome into experience for the next piece of work. It grows out of my earlier Maestro work and points towards AI as a continuing peer that grows with people.
Education
MTech, Hydraulics EngineeringIIT Roorkee · 2001
B.Tech, Civil EngineeringNIT Calicut · 1997
Two current settings
AI Initiative is my independent lab. At Fifth9, I architect systems for intelligence working together. Both inform my perspective, with each organisation’s work clearly attributed.
Founder & AI Scientist
An independent home for a larger vision: intelligence that grows with people. The work includes modular learning, poker and Matiz, games and writing.
Games are the current route to market. The wider direction remains open to other ways evolving intelligence could help people learn, create and act.
Principal Data Scientist
Based at HSBC · Developer Experience
I architect Conduir, an agentic, multi-agent AI architecture in which people and AI agents work together. It composes models and tools for each task, keeps memory that outlasts any single model (Cortex), maps how complex systems fit together (GraFOSa), explores ideas through many perspectives (Manthan), and turns each reviewed outcome into experience for the next piece of work. It grows out of my earlier Maestro work and points towards AI as a continuing peer that grows with people.
Conduir Spatial adds shared environments. The scope and development of each contribution are described in the work map.
The habit running through the journey
I care about ambitious questions and the patient work of testing them. Progress becomes useful when another person can understand the reasoning, question the result and carry the learning forward.
How the work is approached ↗Make the purpose and the question clear enough to examine.
Look at the work, invite another perspective and take unexpected results seriously.
Keep the useful lesson and give the next attempt a better starting point.