Visualization of the datasets codebase

[ivanmilevtues]

Hi all,

I am Ivan and i just finished my masters in AI/ML. During that I've used huggingface at multple occasions so that is also why I decided to get in touch with this community.

Me and a friend are building something to help people get up-to-speed with new codebases. We thing visualizations are the way to go and in this case that means diagrams. We use static analysis and LLMs to generated high-level diagrams, so that people can grasph which are the main components of the codebase and how they interact with each other. So I generated one for the datasets project: https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/datasets/on_boarding.md

I would love to hear what is your opinion on the diagram and what do you think about diagram driven documentation!

Full disclosure: we're trying to turn this into a startup, but we're still in a very early stage and figuring out what will actually be useful for people.

[production23]

Hey Ivan — love the concept you and your teammate are working on. Diagram-driven onboarding is a really creative solution to a problem a lot of devs face.

I’m with a team (EvoLearns) that’s building out tooling for AI workflows — especially around review pipelines, validation, and dataset labeling. We’re currently offering a free pilot program to early-stage teams working with LLMs or ML pipelines.

If you’re interested, we’d be happy to help validate your diagram generation (e.g., reviewing components extracted by LLMs, checking for alignment with repo structure, annotating relations, etc.), or even support any internal dataset creation if you're training your own models.

No pressure — just offering support if it could help accelerate your next steps. Happy to connect if useful!

ir@evolearns.com