Dear,
I laready asked the question on Quora but I feel like there will be intersting advices here. I can’t decide between several notebooks based on the Anaconda distribution (all included) or a Jupyter Hub (not easy to deploy but easy to scale to more users).
Hello! 
Have you heard of “The Littlest JupyterHub” (tljh.jupyter.org)? It’s a (non-Docker) single-VM deployment of a JupyterHub that is much easier to set up, tear down, etc.
Otherwise, if you want to use Docker then the next best bet is probably Zero to JupyterHub for Kubernetes (z2jh.jupyter.org), which is much more scalable, but may be overkill if you just want this for a small team.
I have a simple docker version in https://collaborating.tuhh.de/cmk3624/minimal-jupyterhub-docker-config with some additional explanation in German at https://insights.tuhh.de/de/blog/tools/2019/07/12/pruefungen-mit-jupyter-hub/ - maybe deepl or some other translation website can help you to read it if necessary?
In addition at https://www.tuhh.de/alt/mls/teaching/e-exams.html there is more information available
Take a look at what I’ve done at github:joequant/bitquant
What I’ve done is to create a docker image that contains jupyter, but it’s connected to webmin and nextcloud as well as dokuwiki, and it’s designed for small workgroups. I use it to run my company.
One thing that I’ve done is to set up the docker image so that in include every python library I can find.