SandeshAcharya

I am a computer engineer, trained at the Institute of Engineering under Tribhuvan University. My degree work ran the whole width of the discipline — soldered hardware at one end, deep learning at the other. The final project was a speech-synthesis system that could speak Nepali in a cloned voice from a short audio sample.

I now hold an administrative post at the Embassy of Nepal in London. People read that as a change of direction. I read it as the same work in an unfamiliar language: a diplomatic mission is a system, with queues and state transitions and validation rules and failure modes — and the failure modes matter enormously, because the person on the other side of the counter is a long way from home.

The engineering habit turns out to be the useful part. I notice where a process is doing by hand what it could be doing once, and I have written the tooling to close that gap before — a bulk lookup against the Inland Revenue database, built because checking records one at a time was the bottleneck. Four languages help more than any framework I know.

Underneath all of it is the science. Astrophysics, fractals, chaos — systems where very simple rules produce behaviour nobody could have predicted from the rules. It is not a coincidence that institutions work the same way. That reading changes how I approach both: look for the invariant, the conserved quantity, the smallest model that still predicts the result.