About

About

  I am Mishan Shah. Head of Engineering at LQ Digital, based in Kathmandu. For the past four years I have been building and running production backend systems, Django APIs, FastAPI services, PostgreSQL, self-managed VPS infrastructure, Docker, nginx. Most of it is private client work. The projects page on my portfolio has what I can share publicly. My job is no longer about writing code. It is about architectural decisions, code review, guiding engineers, and figuring out where AI genuinely fits into a product versus where it is noise.

The question

In 2024 I built an LSTM model on Nepal’s stock market. During testing, 33 of the 50 features I trained on were not available at inference time. I padded them with zeros and ran it anyway. The model did not hesitate. It kept outputting confident predictions as if nothing was missing. A junior intern in their first week would have said: I do not have enough information. The model just kept going. That gap has been hammering me since. It is what this blog is about.

This blog

Backend engineering, server infrastructure, and the hard questions about why AI works the way it does and where it fails. I also stream the thinking process live on YouTube at @Mishanrajshah.

  If something here resonates, reach out.