I built a Retrieval-Augmented Generation question-answering pipeline that handles document ingestion, chunking, embeddings, and retrieval. Its modular design separates ingestion, configuration, retrieval, and generation so each component can be tested and scaled independently.
As a Machine Learning Intern at Cybernaut, I built and evaluated Python models on real-world datasets, working across preprocessing, feature engineering, model selection, and performance evaluation. I also designed end-to-end ML workflows around business problem statements.
I developed a multi-agent resume and interview preparation application with four specialized agents, and implemented semantic resume search using sentence-transformer embeddings. My projects also include a PyTorch image segmentation model and an NLP email spam classifier; I presented original research on AI-driven technologies at an international conference hosted by Bharath Niketan Engineering College.
