If your goal is to finish something you started, then pick the
easiest stuff to learn that can produce the
fastest results based on whatever you already know, even if what you know is "out of fashion".
Also, keep in mind that any tool you build can have very modest beginnings. It might not even have a GUI for your first release.
For example, I absolutely hated the Stripe Admin interface when it related to debugging stuff related to another project I was working on, so I wrote a really mickey mouse Node script to pull data from the Stripe API to make it easier to go through the data I needed while I was testing.
That script quickly turned into a simple Electron application that slapped a GUI around my original script. I hadn't really used Electron before and it gave me an excuse to learn it quickly (and reuse some existing UI code I wrote for other projects).
That Electron app then turned into a bigger Electron application that could access multiple APIs and data sources to collect data into local files and databases.
So if you can just pick one small thing that annoys you (I find that something annoying produces faster results than something interesting, ymmv), you can build something modest to address the immediate problem, and then iterate on it to build it into something bigger and better.
I could have just as easily turned my original script into a command line executable, an NPM or a small Dockerized web app container - do whatever is interesting, fun and useful for you.