Building the backend
and the model.

I picked Python as the main thing to get good at, and I split my time evenly between two directions on purpose — FastAPI on the backend and data/ML on the other side — because I don't want to be a developer who's scared of a model, or a model tinkerer who can't ship an API.

Before any of this was coursework, it was curiosity: quantization, LoRA, GGUF files, running models locally with Ollama, poking at how a small LLM gets trained from scratch. That head start is why the ML side isn't intimidating — it's familiar ground I'm now learning to formalize properly.

The goal is concrete, not vague ambition: an internship at a Kathmandu or Lalitpur-based team — Leapfrog, Fusemachines, Cotiviti, CloudFactory, Locus — within six months of starting from zero.

I want to be genuinely useful on a real engineering team within the year — not just able to follow a tutorial, but able to ship something a team relies on. Kathmandu's tech scene is small enough that showing up consistently and building in public actually gets noticed, and that's the bet I'm making.