About

I’m a full-stack engineer in Kathmandu. I build web products for clients in Nepal and France, owning the whole path: the interface, the API and data model behind it, and the servers it runs on. I care most about the parts nobody sees until they fail: authentication, data integrity, deploys and backups.

I’m also working through the mathematics of machine learning, with the aim of building AI into real products properly rather than as a demo.

I started building for the web in 2023 with HTML, CSS and React, and moved steadily down the stack: TypeScript and Node, then Python, Django and PostgreSQL, then the Docker, Nginx and CI work that decides whether software survives contact with production.

Today most of my work is complete systems for small organisations that need them to be dependable: a trekking agency that takes bookings against real departure capacity, a charity that publishes in two languages, a mountain guide whose site had to feel like the mountain.

How I work#

I start from the data and the failure modes, not the screens. What has to stay true (a departure can’t be overbooked, a staff account can’t become a superuser, a backup has to restore) decides the model, and the interface follows from it.

I test against the real thing. Several of the most serious defects in my projects only appeared against production images, a real PostgreSQL and an HTTPS proxy, so that is where the release checks run. When something fails, I write down why, in the commit, the code comment or a report, so the next person, usually me, doesn’t rediscover it.

What I’m interested in#

Backend architecture and the edges between services. Authentication and session security. Performance that comes from rendering less rather than from tricks. Interfaces that stay usable with a keyboard, on a slow phone, in French. Developer tooling that turns a hard-won fix into a default, which is why I extracted my own full-stack starter.

Where I’m heading#

I’m progressively moving toward AI and machine-learning engineering. I’m not there yet, and I’d rather show the work than claim the title. I’m working through a 100-day programme on the mathematics of ML: linear algebra first, then calculus, probability and optimisation, implemented in NumPy with no ML libraries. I’m also studying computer-science foundations alongside it.

The goal is to bring the same engineering discipline to intelligent features that I bring to web systems: measurable, tested, and honest about what they can’t do.

Now

Updated 6 November 2026

Tools I reach for

Interface

  • TypeScript
  • React
  • Next.js
  • Tailwind CSS
  • Three.js
  • GSAP

Backend

  • Python
  • Django
  • Django REST Framework
  • Celery
  • Express

Data

  • PostgreSQL
  • Redis

Infrastructure

  • Docker
  • Nginx
  • GitHub Actions
  • Linux
  • Cloudflare R2

Quality

  • pytest
  • Playwright
  • Vitest
  • Ruff
  • ESLint

Learning

  • NumPy
  • Jupyter
  • linear algebra

Selected experience

  1. 2026–Now

    100 Days of Math for AI

    A public programme on the mathematics of machine learning, implemented without ML libraries.

  2. 2026–Now

    Production platforms for Au Cœur du Népal and LWNF

    Two Next.js and Django systems taken through security review, production validation and deployment, then distilled into a reusable starter.

  3. 2026–Now

    3D expedition site for Sajjan Ghale

    WebGL terrain from survey data, scroll-driven camera, adaptive quality.

  4. 2025

    Django, REST APIs and data work in Python

    Django and Django REST Framework, NestJS, data science with Python, and data structures and algorithms.

  5. 2024

    First client project: Lucky Dairy

    A React and TypeScript front end with an Express API for a dairy business in Ilam.

  6. 2024

    TypeScript, Node and Next.js

    Moving from interfaces to servers: TypeScript everywhere, Express, and the first Next.js applications.

  7. 2023

    Started building for the web

    HTML, CSS, Bootstrap and React, alongside C, Java and Python fundamentals.

esc