I’m working on dashboards and scheduling for Jupyter Notebooks. So far, I’ve created the proof-of-concept repo https://github.com/mljar/mljar-studio
The main trick is to store cell’s output in the database and have separate webserver to make it available (without connecting to Jupyter server).
In my solution I’ve created my own notebook front-end which is a drawback of my solution because I think it will be much better to integrate with existing front-ends (Lab, Notebook or nteract). Do you think is it possible to create such extension that will synchornize notebook frontend with external server that will store information about cell and cell’s output?
2 Likes
Could you elaborate on what you mean with scheduling?
You can put your Notebook in the scheduler for periodic runs, for example run notebook every 1 hour or run every day at 8:00 AM. In the dashboard you will see outputs with updated values.
1 Like
Hey @pplonski I have a project where we build/schedule/run notebooks using repo2docker. We don’t stream cell outputs directly but serve saved files from an API which a frontend dashboard can read.
We currently run it as a managed service but want to make it portable
1 Like
Hi guys,
Following the subject,
we have made scheduling very easy in with our opensource layer on top of Jupyter : https://www.naas.ai/
Only with : naas.scheduler(“a_cron_task”)
I would love to show you a demo if any of you guys have some time.
We have it deployed on Kubernetes/AWS but you can run on your local as well.
Best,
J. from Paris 
1 Like
@Jeremy_Ravenel, do you have the Terraform code, or any tutorial about deploying Naas on AWS/Kubernetes (EKS I suppose), thanks!