Raj Palleti

I'm a Robotics/NLP researcher at Stanford University currently working on robotic manipulation via language and human-robot interaction. I previously did research at Stanford ML Group and Stanford Hospital working with Percy Liang, Dorsa Sadigh, and Jure Leskovec on various projects spanning healthcare ML, supply chain optimization, and language-guided robotics.

My work focuses on making robots more adaptable and intuitive to interact with through natural language. I'm particularly interested in shared autonomy approaches that combine human input with learned models to enable more reliable and sample-efficient robot learning. A key focus has been developing methods for robots to understand and incorporate natural language corrections during task execution.

Beyond robotics, I work on applications of machine learning in healthcare, including using deep learning and transfer learning for ECG analysis and risk prediction. I'm also interested in supply chain optimization, where I've worked on using graph neural networks to model production functions and forecast transactions between firms.