Hello NVIDIA team,
I am Hari Dangi from Nepal. I started learning AI when I was around 14, and now I am 20. Since then I have been learning and building mostly from open-source projects, papers, docs, and my own small computer.
I am writing here because I do not know the right NVIDIA person or program for this, and I wanted to ask directly. I need help with one specific thing: one DGX Spark.
I am not asking for cloud credits or a large server. Cloud credits expire and I cannot afford cloud long term. A bigger machine will be hard for me because of electricity and setup cost. A smaller machine will still not be enough for the kind of AI experiments I need to run. So DGX Spark feels like the exact middle ground for me, small enough to keep locally but powerful enough to validate my research.
For the last several months I have been working on Keypra, which is a long term memory AI system for education and learning, especially for students and teachers in Nepal and remote places where internet is weak, resources are limited, and education support is not enough.
I am also working on Cactus, an AI mobile agent, and doing my own research on long term memory and persistence AI systems. I am trying to build something useful not just another slop thing or personal toy.
The main obstacle stopping me is compute, constant compute. I can read papers, write code, understand code, think, design, and build software, but my current machine is too limited for long term testing, local inference, agent experiments, and properly validating memory systems.
Right now I am using an general laptopIt is ok for normal work, but for AI research and local model testing it is very limited. It also does not come with the NVIDIA stack and convenience that DGX Spark has.
I come from a lower-income family in rural Nepal. My father earns around $4,000 a year and most of that goes into family survival. I call it survival cause it is survival. A normal day of work here can be around five dollars, so buying DGX Spark myself is not realistic right now.
I do not want to present myself as helpless. I am not helpless, I am building. But I am constrained by the machine in front of me.
If NVIDIA can support me with one DGX Spark, I will use it seriously for Keypra, Cactus, local AI education systems, long term memory research, and agent experiments. It would not be wasted. It would directly become the machine I use every day to build and test.
If there is any NVIDIA person, developer relations team, creator team, regional team, or program that can review this kind of request, please point me there or forward this. or email me .
I know this is not the right place to ask, but I have no other way, and I am feeling kind of embarrassed.