I am thinking of using Ray as the compute backend for my internal machine learning platform. But I am not sure what I can uses as a frontend. I need a landing page UI to link to notebooks, experiment tracker (MLFlow), Tensoboards, pipelines, etc. but I do not want to build a custom UI from scratch, as our team is all backend engineers. If anyone has used some OSS framework successfully for this frontend purpose, and can share your experience, it will be a huge help.
Basically an OSS alternative to Anyscale UI.
I am curious if you checked out Gradio. Ray Serve provides a simple integration.
Gradio has integration with weight and biases.
That would be a really cool application. Very curious with what you come up with.
That is an interesting thought - host a central dashboard UI using Ray serve. I have not thought about that, but I will look into it.
If we can keep the central dashboard simple, it probably makes sense to just keep it static and hold links to all other dashboards. But Gradio or Streamlit type of solution might be a good stepping stone to get a kick start.