Compute
Who supplies it, who can withdraw it, what can run locally, and which workloads require assured capacity?
I write about the questions between policy and code: who controls compute and data, where authority sits in an AI system, how infrastructure dependencies become strategic dependencies, and what India, the UK, the EU, and the US can operate for themselves.
I completed the Graduate Certificate in Strategic Studies at The Takshashila Institution in 2018, alongside a course on Xi's China and a Cogitatum workshop on bringing fugitives back to India.
Strategic studies added a second scale of analysis to my engineering work. A runtime policy, model dependency, cloud region, signing key, or data boundary is also a distribution of authority between people and institutions.
That perspective now informs work on governed agents, local inference, regulated financial-services AI, blockchain signing policy, and cloud-to-edge systems. It also provides a way to examine technology sovereignty without treating domestic ownership, open source, data residency, operational control, and supply-chain resilience as the same thing.
The Takshashila Institution ↗Who supplies it, who can withdraw it, what can run locally, and which workloads require assured capacity?
Can the organisation inspect, replace, evaluate, and operate the model and runtime under its own constraints?
Which jurisdiction, institution, and accountable person can authorise collection, movement, retention, and reuse?
Who can approve an action, interrupt it, reconcile failure, and remain responsible after automation?
Policy, market, programme, and technology details reflect the date shown on each essay and may have changed since publication.
Available for talks, podcasts, panels, and briefings on AI sovereignty, runtime governance, technology dependencies, and the institutional design of AI systems.
Speaking topics