same here - actually, today we use zeppelin, which is kind of a pain to setup on a cluster.
I see that you are using Reactjs. If you can build something that can connect to spark and python, I will pay for this and so will lots of people.
One of the big unsolved issues where we want to really throw money is have a notebook that my analysts can work on and then I can take straight to production and have it run multi-user.
There are lots of startups attempting to do this on top of Jupyter, but I believe building this from scratch is the right way.
I'm literally crying right now for a reactjs + d3js dashboard running on pyspark that multiple people can use simultaneously.
Analytics like sum(column) are much faster if data is saved column-oriented. Also every time you don't do select *, but select a, b, it's much easier on disk because you're not even touching some columns.
Currently not but since the preview is just normal HTML, all we <technically> have to do is load a different css file. You could of course also change the shipped css file and build your own version.