That's a platform designed for journalists that indeed contains a lot of scraped data and leaks. It cross-references them, so you can type in a name of the person/company and see the data points mentioning them. Some of the investigations got started precisely because journalists were able to use it to see a connection you can't see otherwise. It handles all the quirkiness that arises from dealing with different alphabets, and it accounts for multiple ways you could type a name from a different alphabet into Latin.
The administrators of it manage who gets to see which data sets for obvious reasons. Knowing nothing about this specific investigation, I assume that the data is currently only accessible to the pre-approved journalists (that work on this investigation).
The code of the tool is available on GitHub, and if you happen to stumble upon a security issue: https://www.occrp.org/en/responsible-disclosure