The mapping between different character sets isn't guaranteed to be one on one in all cases; that is, not every glyph/letter sequence in one language maps to a valid sequence in another language. Multiple different sequences in one set might even map to identical sequences in another. Therefore such a functionality can not be automatic, you probably won't be able to enter a Hindi character sequence on signup and have it automatically create an account with a matching english name, which then also maps to arabic, hebrew or greek.
Allowing the user to create additional login names in different character sets for their account might be an alternative, but this might lead to security problems by adding more valid login/password combinations, thus reducing the search space, or by creating ambiguities (different users suddenly having the same login).
But using different character sets might create other problems: what happens when the password for such a login is entered? Must/can it be given in the same character set as the login, or in a different character set? How does the user know which character set is used when he types in his password? How would password rules be enforced?