EkNepal: A First Attempt at Building Around Uncertainty

Could you get truthful information?

Open a handful of Nepali news sites in a row and you get a handful of different accounts of the same country — the same event described differently, or reported by one outlet and missed entirely by the next. Nothing was necessarily wrong; there was just no single place that put the coverage side by side and let a reader compare it. That gap — not a bug, not a feature request, just a genuinely fuzzy problem — was the starting point for EkNepal.

EkNepal is a Nepali-language news reader that pulls reporting from several distinct outlets — among them Himalaya Times and News of Nepal — into one continuously updating feed, each item clearly attributed to its source publication. Articles carry share links (email, Facebook, X, Viber, WhatsApp) and a comment section, and the site is built mobile-first, with dates rendered in the Bikram Sambat calendar alongside a live weather readout. Beyond news, it bundles a set of everyday utilities Nepali readers commonly look for alongside their news — foreign-exchange rates, share-market figures, a horoscope, a date converter, a Unicode converter, and gold and silver pricing — organised behind a single navigation drawer.

EkNepal is a product built and maintained by a team; it is not something I built alone, and I did not build its backend, its news-ingestion pipeline, or any fact-checking or source-classification system.

I contributed to the frontend of EkNepal through Alpinist Studios — the part that turned scattered reporting from multiple outlets into something a reader could actually navigate, read, and compare. It was my first paid software engineering work, and the first time code I wrote shipped to real users rather than a course instructor.

The habit EkNepal left me with wasn't a framework or a library — it was a way of starting: begin with a question that doesn't have a clean shape yet, break it into pieces small enough to actually build, ship something testable, and see what survives contact with real use. That's the same loop I now run on algorithms and proofs — state the conjecture, try small cases, build the smallest thing that could be wrong, and let the evidence tell you where the structure actually is.