This study addresses the limitations of conventional meal search engines in fitness applications by introducing a comprehensive approach that integrates semantic search techniques with state-of-the-art Sentence-BERT models. Traditional keyword-based search systems often fall short of capturing the context nuances and user preferences associated with meal queries. Hence, this research proposes an enhancement that employs advanced Natural Language Processing (NLP) and Deep Learning (DL) methodologies. It leveraged the SBERT to embed the semantic meaning of meal-related queries, allowing the search engine to understand user intent and preferences. In addition, this research introduces a practical implementation, featuring the development of a mobile application with a fully functional backend server to demonstrate the proposed enhancements. This research contributes to the ongoing efforts in advancing mobile application technologies, providing a foundation for improved meal discovery experiences in the rapidly evolving digital landscape.