Noorin Nasir Khot

Multimodal AI Document Intelligence Platform

Description: Developed a multimodal RAG system that processes PDFs, scanned documents, and reports using OCR, layout understanding, and hybrid retrieval to answer natural language questions with accurate, citation backed responses. ◦ Technologies: Python, LLMs, RAG, LangChain, OCR, Hugging Face Transformers, Vector Embeddings

Image Captioning with Transformers

Description: Designed a deep learning model that generates text captions for images using a CNN encoder and Transformer based decoder with attention. ◦ Technologies: PyTorch, ResNet, Transformer, Attention

Sentiment Analysis using Deep Learning Models

Description: Implemented and analyzed multiple deep learning architectures Vanilla RNN, LSTM, BiLSTM, GRU, and BiGRU for binary sentiment classification on the IMDb dataset. ◦ Technologies: Python, Keras, RNN, LSTM, BiLSTM, GRU, BiGRU, Streamlit, NumPy, Matplotlib