Rajeev Ambati

I am a second-year PhD candidate in the UNC-NLP lab, advised by Shashank Srivastava and Snigdha Chaturvedi

My research is interactive learning, where we explore how intelligent systems can learn continuously through interaction with humans. I develop methods that enable AI agents to identify gaps in their knowledge, actively seek information through natural language, and efficiently incorporate newly acquired knowledge.

Before joining UNC, I spent a few years in the industry, where I developed and trained deep-learning models for healthcare, finance, and camera systems. Prior to that, I got my B.Tech from IIT Kanpur and M.S from Penn State.

UNC Chapel Hill

Research Interests

My work centers around the following research questions:

How can machine learning models learn from interaction via natural language?

How can models reliably recognize what they do not know?
How can they ask informative questions that maximize knowledge acquisition while minimizing interaction?
How can newly acquired knowledge be retained for short-term tasks and long-term continual learning?
What theoretical guarantees can we provide regarding learning efficiency and generalization?

It has applications in education, healthcare, robotics, and other domains where AI systems must continually adapt to evolving environments. As AI becomes increasingly integrated into society, I believe that interactive machine learning will play a central role in building robust, reliable, and lifelong learning agents.