Hi, Oleg!
Since you’re facing the challenge of providing multilingual access to data, maybe you would like to take a look at the pattern for language support in Frictionless Data, which provides a couple of ways to use multiple languages both in data and data descriptors (metadata). The discussion about it is still open, with a settlement about the proposals scheduled for the v1.1 of the Data Package specs.
I myself find this to be an interesting challenge. Open Knowledge Brasil hosts project Serenata de Amor, a famous project that uses data analytics and machine learning on open data to find suspicious uses of reimbursement of expenses by members of parliament (note: in the EU this kind of data is kept secret). One of the most heated debates in this project has been whether to use English or Portuguese on its Github repositories and Telegram channel. Using Portuguese is best for getting engagement from local activists, which are likely more motivated to contribute to the project. Using English, on the other hand, gives the project more international visibility and recognition – and occasional skilled contributors.
My opinion is that it’s best to be multilingual as you get the best of both worlds. I try to post bilingual content on my blog, even if I take longer to write and can’t post as often. The same with some of my more popular Github repos. I understand that it can be burdensome to feature more languages, and keep content in sync between them, but I think it’s worth it in the end.
Ausgezeichnet! 
You are you planning to do this remote participation? Should we register for that kind of participation as well?