Hello, I'm Shashank Sonkar, currently a PhD student at Rice University advised by Dr. Richard G. Baraniuk, with a specialization in Natural Language Processing (NLP) for personalized education. My primary research focus lies in understanding, training, and optimizing Large Language Models (LLMs) to unleash their potential in educational settings. Please find my publications list here.
I possess extensive expertise in training LLMs for education tutoring to enable personalized, engaging, and effective learning experiences for students. My research also delves into understanding the limitations of LLMs in educational settings. By examining the potential pitfalls of deploying LLMs in the education domain, I strive to address critical issues such as robust student simulation using LLMs to enhance the overall efficacy of LLMs in education.
My past projects revolved around learning knowledge graphs and understanding why deep NLP architectures work, such as the roles of attention mechanisms and feed-forward networks in parallel transformer architecture.
Before joining Rice University in the fall of 2017, I worked as Senior Research Scientist at Media.net, Mumbai, starting from 2013. At Media.net, I worked on designing algorithms that power “Yahoo! Bing Network Highly Contextual Ads Program“ using Deep Neural Networks. The algorithm intelligently mined for commercially and semantically relevant ad-keywords for a URL.
I earned my Bachelor's degree in Computer Science & Engineering from Indian Institute of Technology, Kanpur (IITK), where I was privileged to conduct research under the expert guidance of Dr. Harish Karnick.
Most of my code and other extra curricular projects is hosted on Github.
Research Interests :
Large Language Models (LLMs) and their applications in education
Understanding inner workings and limitations of LLMs
Learning and Reasoning over Knowledge Graphs
Email : sonkar.iitk@gmail.com, ss164@rice.edu