Long before I taught machines to see, geometry taught me to look. It is the oldest of the sciences, and still the sharpest instrument we have for cutting complexity down to structure.
My doctoral work in computer vision at Queen Mary University of London lived in the hardest corner of that discipline: years spent staring at surfaces that gave nothing back. No edges, no texture, no helpful contrast; scenes a person reads at a glance and a machine cannot read at all. I learned to stop asking what things look like and start asking what they are: where appearance fails, structure is all that remains.
It turns out operating theatres are not so different.
Surgery is visually ambiguous in exactly that way. Critical signals are subtle; errors are rarely dramatic in the moment, and their consequences almost always are.
The Investment
None of it was self-made. I studied mathematics at the Indian Institute of Technology Madras after securing a top national rank, and every step since has been carried by scholarship and support.
After IIT, I spent six years as a Lecturer in Mathematics at Alfaisal University in Saudi Arabia: six years of teaching, learning to make hard ideas hold up in front of a room, and earning my first research grants in computer vision.
Then Queen Mary University of London awarded me a full Principal’s Postgraduate Research Studentship for doctoral research in computer vision: tuition, living, three years. Everything.
Without that support, I would not be here.
A debt like that is not repaid to the people who gave it. It is paid forward, at scale. For me, that means applying mathematics and engineering where they improve real systems, and few systems touch more lives than surgery, millions of times a year.
The Turn
I grew up in India, where I saw what healthcare looks like when the infrastructure behind it fails. Benchmarks stopped feeling like the point.
That conviction is why, in 2018, I co-founded Scalpel AI, where I serve as CTO.
After ten years in academia, I left. Not because the research stopped mattering, but because the operating theatre is where it matters most.
The Work
The same discipline, a new arena
The discipline transferred whole. A mathematician’s job is to capture the structure of a scene in models that can be trusted. At Scalpel AI, the scene is the surgical pathway, and the models carry real consequences.
We are building the infrastructure of Connected Surgical Intelligence: systems that watch the whole pathway, from the warehouse that packs a tray to the theatre that uses it, and verify each step as it happens. The work rests on three pillars.
01
Instrument Intelligence
Automated surgical instrument detection and tray verification: identity read instrument by instrument, wherever the tray happens to be.
This reduces the risk of retained foreign objects, strengthens compliance, and introduces accountability beyond manual processes.
02
Workflow Intelligence
Real-time analysis of surgical workflows. Computer vision, machine learning and contextual signals read the sequence of steps and verify adherence to protocol.
Invisible deviations become observable data.
03
Operational Intelligence
Surgery is also a logistical system. We give inventory, tray utilisation and workflow coordination the same data-driven visibility.
Clinical and operational teams work from one picture, with continuous feedback across the system.
Connected Surgical Intelligence does not add another tool to the theatre: it connects the ones already there.
Leading that work is its own discipline. I build teams that value precision over hype and robustness over shortcuts, because in systems this close to patients, elegance is not enough: they must hold up under constraint, uncertainty and scrutiny.
Models evolve and tools improve. The architecture underneath must stay coherent, interpretable and defensible. That is the standard the team signs up to.
Connected Surgical Intelligence is a technical challenge. But it is also a leadership challenge: building it properly requires patience, discipline, and a commitment to earning trust over time.
The work has been backed at every stage by competitive funding, most recently multiple Innovate UK awards at Scalpel AI. The full ledger is on the Record page.
The Foundation
Where this is heading
We are only at the beginning of making surgery measurable.
The hardest problems in surgery are not purely technical. They are systemic: workflow gaps, communication breakdowns, risk that accumulates quietly until it becomes visible too late.
They are not solved by adding isolated tools. They need connected systems that understand context, sequence, and consequence.
Connected Surgical Intelligence is the foundation the tools stand on.
A foundation where the theatre, the workflow and the supply line share one picture, and nothing in a patient’s care depends on what somebody happens to remember.
Mathematics gave us the tools to model structure.
Computer vision enabled machines to see.
Now we must engineer systems that can understand.
Isolated AI models will not define the next decade of surgery. The infrastructure that connects them will.
That is the foundation I am building.