In theory, if Google supports Tamil (it does) as an interface language, then its algorithms should also operate in that language. Whether they are as "perfect" as they are in English is a different matter. For example, non-English languages don't sort as per the English alphabet. Plurals often don't exist. Human raters might not be involved. Spam might not exist, or it might have a different fingerprint. etc etc
I don't know Tamil but I know Gujarati, and a search reveals most of the top results have their meta tags in Gujarati.
In your case, you need to ask yourself whether searchers will search in English or in Tamil. I feel that someone searching in English is unlikely to be shown a Tamil result if there are plenty of relevant English pages. If your audience is likely to have their default search language set to Tamil (not available in Google, even though the interface language can be set to Tamil), they will probably see Tamil results.
In conclusion, since Google has a large number of interface languages but a small set of search languages, I feel that the algorithms in unsupported languages are basic. I copied one of the Tamil words second from the top left nav bar (presume it was Images) from the Google page and when I searched, I was given mainly Tamil results.