Kathmandu, Nepal
I'm a student of computer science who spends most of my time thinking about data — how to understand it, how to learn from it, and how to turn it into something useful.
01 — Personal
I was born in Dang Ghorahi, a valley in the midwestern hills of Nepal — wide fields, slow rivers, and the kind of quiet that only a small town can hold. I loved growing up there. The place taught me to pay attention to small things, and I carry that with me.
I came to computer science sideways. I was curious about how systems work, how patterns repeat, how a small observation can explain something much bigger. Data felt like a language for that kind of curiosity.
Outside of building things, I read slowly, walk often, and take photographs of places that feel like they are holding still. I care more about understanding than impressing, and I try to keep things honest.
Things I care about
Essays, philosophy, and the occasional novel — anything that slows me down.
Old streets, morning light, and the quiet moments people walk past.
The best way I have found to think through a problem.
Helping friends with concepts I once struggled with myself.
Eventually, curiosity became work.
What started as a desire to understand things turned into a practice of building things.
02 — Professional
I work with data and machine learning. I build systems that help people understand information, make decisions, and notice patterns they would miss on their own. Most of my work sits somewhere between analysis and engineering — taking a real, messy problem and shaping it into something a model or a dashboard can actually work with. I care about clarity. I would rather build something small that works well than something large that no one understands.
Right now I'm working deeper in computer vision and applied deep learning — moving from using models to understanding how they work, and occasionally how to improve them.
Experience
Working with datasets that describe real behaviour, and trying to ask better questions of them.
What I worked on
Building predictive models, cleaning data pipelines, and creating visualizations that people can actually use to make decisions.
What I learned
Most of the work is asking the right question. A good model on the wrong question is worse than a simple answer to the right one.
Contributing to machine learning libraries and research projects in my own time.
What I worked on
Reviewing implementations, reproducing papers, and writing small tools that make experiments easier to run.
What I learned
Reading other people's code teaches you more than writing your own. Open source is a long, patient conversation.
Selected work
A comprehensive analytics dashboard built to take large, unstructured datasets and turn them into something a person can read at a glance — with predictive models running underneath to flag what is likely to happen next.
Why I built it
I wanted to understand what happens after the model. A prediction is useless if no one can see it or act on it.
What I learned
The interface matters as much as the algorithm. A good model behind a confusing dashboard changes nothing.
Computer Vision
Real-time object detection for images and video streams.
I wanted to see how a model perceives the world in real time — not on a benchmark, but on a live camera feed.
Code
AI Applications
An NLP tool that reads text and surfaces the emotional tone underneath.
Language carries more than information. I wanted to see whether a model could pick up on the texture of what people say.
CodeCapabilities
Next.js, React, TypeScript, Django, REST APIs
Python, SQL, PostgreSQL, Pandas, Power BI, Tableau
Scikit-Learn, TensorFlow, PyTorch, OpenCV, YOLO, NLP
Git, Docker, Linux, Google Colab, Jupyter
Currently exploring
Going deeper into how transformer models actually work — not just using them, but reading the papers and re-implementing parts from scratch.
Applied computer vision for real-world, messy environments — not benchmarks, but actual cameras and actual light.
Writing more about what I learn. I have found that explaining things clearly is the fastest way to find out what I do not understand.
Contact
I read every message. The best way to reach me is by email, but I am around on the usual places too.