Yamini Singh

I'm building AI systems across document intelligence, healthcare, and enterprise engineering workflows. At STMicroelectronics, I developed RAG pipelines with LangChain for retrieval, reasoning, structured data processing, and contextual decision-making. I also supported the People-Services eLeave platform using Java, JDBC, HTML, CSS, JavaScript, XML, and REST APIs, while contributing to PeopleSoft-to-Oracle HCM Cloud data migration. My machine learning work includes a Keras churn model that achieved 79% test accuracy and deep learning research at CSIR on medical imaging, GANs, and knee osteoarthritis prediction, improving model performance to 88%. I've built a rental-agreement metadata extraction system using PaddleOCR and Gemini, an AI-powered smart cane with 86.2% accuracy, and Fashion GANs in PyTorch. I also co-authored an IGI Global book chapter on assistive navigation for visually impaired persons.