This research proposes a hybrid document retrieval system that combines keyword-based and semantic search techniques to address the limitations of traditional methods. While keyword-based systems (e.g., inverted indexes) excel at exact matching, they lack semantic understanding. Conversely, semantic methods (e.g., word embeddings) capture context but often miss precise query terms. To bridge this gap, the system integrates GloVe embeddings for semantic representation and FAISS for efficient similarity searches, alongside an inverted index for exact keyword matching. This dual approach balances speed and accuracy, enabling effective handling of multi-format datasets (PDF, Word, Excel, HTML). The framework is designed for precision and scalability domains, such as healthcare, education, and law. Evaluated using Precision@K, Recall@K, and MRR, the hybrid system demonstrates improved retrieval speed, relevance, and scalability, offering a robust solution for modern information retrieval challenges.