Daniel Sitnik

While both are great and there are some web application frameworks for Python, I still see it shining a lot more in data science applications.

That being said, I recommend Node.js. You will be using the same language you’re already working with on the frontend, there’s a large community producing lots of content and tutorials, and also an endless choice of frameworks and tools to choose from.

If your first goal is to learn the basics of developing a web app composed of a frontend that communicates with a backend through REST APIs, then I would go with the Angular + NestJS combo.

My reasoning is that they both use TypeScript (a superset of JavaScript) and the architecture of both frameworks follow the same patterns. Both applications (front and back) will be structured around Modules, Controllers and Services. You need to learn just one language, and since the frameworks are so similar, the concepts you learn from Angular can easily be related to Nest and vice-versa.

Once you learn the basics you can start switching frameworks and languages and going for something different.

Metabase is more user friendly and totally targeted at non technical people. Setup is super easy and it supports all of the most popular datasources.

You can give meaningful names and descriptions to your tables and columns. This helps the end users when they are creating the queries (called "questions" in Metabase). Question creation is all done visually, you can even join tables and do aggregations (group by) without coding any line of SQL, although SQL can be used for more complex cases.

If this data is being sent to a centralized database and the charts and diagrams you need to show are not very complicated, then I would go for something more like a BI solution.

These tools can connect directly to your database (and many other types of datasources) and they offer you a user friendly interface where you can build the queries to extract the data you need and show it in a variety of different charts.

There are some great free and open source alternatives like Metabase and Superset, and of course the commercial ones like Retool and PowerBI.

I would only go the React way if you need something very customized to your needs. Not only you're gonna have to build the full React frontend, you're also gonna need to build the backend to extract the data and provide the APIs for the frontend.