Harsh Vardhan
Research Scientist
I completed my B. Tech in Electrical and Electronics Engineering in 2009. During my undergraduate, I won multiple national and international robotics competitions that ignited my interest in embedded programming and system design. I worked on different position in India in heavy engineering especially in energy sector from 2010-2017. In 2018, I embarked on a new chapter in my academic pursuits by joining Vanderbilt University. Over the course of my tenure, I diligently pursued and attained both a Master of Science (M.Sc.) and a Doctor of Philosophy (Ph.D.) degree in computer science, culminating in my graduation in July 2023. Throughout my research tenure at Vanderbilt, I was affiliated with the Institute for Software Integrated Systems, worked with Professor Janos Sztipanovits. My research endeavors received substantial support from prestigious entities, including DARPA through various programs, the National Science Foundation, and NIST. This robust backing underscores the profound significance and potential impact of our collective research efforts.
Areas of Expertise
Artificial Intelligence, System design, Machine Learning, Reinforcement learning,Applied Mathematics, Optimization, Computational Fluid Dynamics, Probabilistic Learning, Bayesian Optimization , Kernel Learning, Transformers, LLMs, Graph learning.
Research Interests
My research interests are multifaceted and encompass the integration of artificial intelligence, applied mathematics, optimization, and machine learning techniques to innovate in the field of cyber security defense, data-driven model discovery, and engineering design. Within the realm of cybersecurity, my dedication lies in pioneering automated cyber defenses. This involves training AI agents, crafting simulation and emulation platforms, and harmonizing various AI methodologies to fortify our digital security. Notably, I actively contribute to the DARPA-sponsored Cyber Agents for Security Testing and Learning Environments (CASTLE) program. In the domain of engineering design, my research is concentrated on enhancing smart electric grids, design of Unmanned Underwater Vehicles (UUVs), and Unmanned Aerial Vehicles (UAVs). I aim to research on AI-driven automation into existing design optimization processes to enhance efficiency and foster innovation. My approach leverages computational physics, simulation processes, scientific computing using numerical methods, and the development of novel algorithms using AI and mathematics. The sub-fields of AI that I work with, encompass Deep Active Learning, Surrogate Modeling, Bayesian optimization, Bayesian experimental design, reinforcement learning, computer vision, transformer-based learning, graphical probabilistic modeling, Deep Learning, and other machine learning methods. My endeavors extend to controlling nonlinear systems through reinforcement learning, detecting outliers and anomalies in streaming data, and generating test cases for rare event failures in AI systems, all with the overarching goal of bolstering the reliability of AI and ML models.