Generated illustration of a robot-arm experiment beside a paper, notebook, overhead camera, and model traces on a laptop.

Gnaroshi

I study AI systemsand build software for research.

Right now, I’m interested in experiment infrastructure for vision-language-action models and better ways to read, revisit, and connect papers.

Now · A shared experiment layout for running VLA models with separate model, method, environment, and result settings.

Research

Questions I’m exploring

  1. What kind of experiment structure makes different VLA models comparable on common ground?
  2. How can computation be reduced without hiding changes in model behavior?
  3. How can AI tools support research without replacing the researcher’s judgment?
More research questions

Project

gnaroshi_vla

How this site works

From private notes to the public site

Papers and writing begin in private repositories. Studio reviews what should be shared, builds a public record, and the website presents it.

From private notes to the public sitePapers and writing begin in private repositories. Studio reviews what should be shared, builds a public record, and the website presents it.1Paper notesgnaroshi-paper-lab1Writinggnaroshi-writing2Gnaroshi Studiognaroshi-studioAI servicegnaroshi-apiOptional3Public contentgnaroshi-content-feed4Websitegnaroshi.github.ioFrom private notes to the public sitePapers and writing begin in private repositories. Studio reviews what should be shared, builds a public record, and the website presents it.1Paper notesgnaroshi-paper-lab1Writinggnaroshi-writing2Gnaroshi Studiognaroshi-studioAI servicegnaroshi-apiOptional3Public contentgnaroshi-content-feed4Websitegnaroshi.github.io
  1. Private sourcesPaper notes · Writing
  2. Review and publishGnaroshi Studio · AI service (Optional)
  3. Public recordPublic content
  4. PresentationWebsite
See the full architecture