This repository implements a deep learning model inspired by pix2code that takes screenshots of UI designs as input and outputs corresponding HTML and CSS markup. The model uses a convolutional neural network to encode the visual input, followed by a recurrent neural network (LSTM or GRU) that decodes the image features into a sequence of domain-specific tokens. The Bootstrap-specific version achieves 97% accuracy and can generalize to new design mock-ups, while a raw HTML version remains experimental.