Not 100% sure what you mean by works, but Nim has been used to create various things. For one the entire Nim forum: https://forum.nim-lang.org/ is written in Nim (both back-end and front-end), the game "Reel Valley" has been written in Nim, there is a company doing Ethereum blockchain using Nim: https://status.im/, and many more.

I used Nim a little while back to write a small CLI program that performs a feature-preserving smoothing of digital terrain data (see https://github.com/jblindsay/fp_denoise). I could have used, and have used, several of languages for this type of application. But I was interested in learning Nim and getting a feel for how well it lends itself to spatial analysis. I was quite pleased with the experience overall. The language feels a lot like a high-level language with the speed of a systems language. The one thing that I didn't find particularly easy was concurrency. I still haven't managed to figure that out but that is more a comment on my limitations than the language. Nonetheless, I can certainly see Nim taking off for data science once it hits v1.0.