Quant here, at my firm we've deployed a JupyterHub server which provides users with a production docker image, so that analysts and portfolio managers can perform analyses without installing python, dependencies and sql drivers locally. It is working well and spurs interest in Python across the wider org - so we let everyone use it.

Similar to in OP's case, I think the real selling point of Python over Excel is advancing the capabilities and the scale of the business. Talks of different programming languages falls on flat ears in finance - show what can be done instead. With Python, Zipline and notebooks I can manage a global equity portfolio, continuously adding active strategies and adapting to real-world changes and constraints. And backtest! Excel is great, but there is an upper bound to what can be reasonably done without a thriving open source community.

That's really interesting. I've been working on a JupyterHub / JupyterLab / Python based product for the insurance industry - choosing this for all the reasons you've cited. Would be really interested if there are any points you can share e.g are you using Kubernetes and if so how have you found it?

We use Kubernetes, according to our IT/devops guys it was pretty straightforward to deploy in Azure with Jupyterhub's KubeSpawner module + documentation. A few people are quite eager to learn Python / code in general, so we try to make it convenient. One common use case would be to work with existing excel spreadsheets, so the notebook volume storage should be mountable in Windows. The file upload/management in notebook servers is quite obtuse IMO.

If a recurring task can be reasonably parameterized then a Streamlit app might be a better choice in some instances. I've developed a monitoring application for our portfolios where I can track daily asset weights, underlying data points, computations etc. Not displaying code ensures that the output can be consumed by a wider audience.

Thanks especially for the Streamlit suggestion.

We've tested JH with K8 in GCP which was straightforward also. With a small team though tending towards a single VM deploy (based on "The Littleist JupyterHub") which looks a lot easier to maintain.

Thank you for sharing.

Is this really an issue for the use case being described? Environment management is obviously a significant consideration for software developers who need to keep track of versions etc, but it sounds as if these users primarily want to use the fundamental numpy functions.

They could install one of the scientific python stacks (e.g. anaconda) or just install packages globally with pip.

Things always break over time. I have a very technical co-worker, a systems admin, with a broken Python stack on Windows. No idea what's wrong.

so he did a chmod -w on their folder and it broke. Well, that's a feat I still have to work out!

Anaconda switched its software license recently to require large companies to purchase a license, so now people must wade through the purchasing department before installation & use.