Nanda Hady Mulya

Why does so much "AI" feel bolted on — instead of like it truly gets it?

We have never had more capable models. Yet most products treat intelligence as a feature to ship, not an understanding to earn. They answer the literal question and miss the human one underneath it.

I am interested in the opposite: software that grasps context, holds onto intent, and responds the way a thoughtful teammate would. That belief shapes everything I build.

To answer it, it helps to know how I work —

I am a software engineer who treats AI as part of the craft — not a layer pasted on at the end.

My foundation is ordinary, honest engineering: clean interfaces, dependable backends, infrastructure that stays up at 3 a.m. On top of that I build with RAG and LLMs, so the intelligence is woven into the product rather than wired around it.

That means I can carry an idea the whole distance — from the first frontend sketch, through the backend and retrieval logic, all the way to deployment and the servers it runs on. One person, one coherent thread, from concept to production.

That range shows up most clearly in the tools I reach for —

But tools only matter in service of results. Here is the proof —