Currently, Wikidata includes a lot of items around people. Most of these items do not have labels in multiple languages and scripts. Using rule-based transliteration systems such as the Universal Script Converter Extension of MediaWiki does not usually solve the problem as it does not cover all the languages and as there are a lot of complex ways of doing transliterations that cannot be supported by rules. With the new hype of LLMs, we believe that this problem can be solved. In this project, we create a transliteration tool for proper names based on a language model hosted on Hugging Face and fine-tuned using a Wikidata-based SPARQL-generated dataset.