Sign language recognition is a critical aspect of enhancing communication accessibility for individuals with hearing impairments. Recognizing the complexity and nuances of sign gestures can be challenging, and automating this process can significantly improve efficiency and accuracy. Leveraging deep learning techniques with frameworks like Tensor Flow, Keras, and Mediapipe, we propose an automated system for sign language recognition. The objective of this system is to expedite the recognition and classification of sign gestures, providing a quick and reliable non-invasive solution for users and facilitating communication. Our approach involves transfer learning using pre-trained models in conjunction with custom architectures tailored for sign language recognition.