My Motivation
I like to explore a lot, both academically and geographically. Consequently, I have a varied technical background and have worked/studied in five countries.
I am attracted to all things mathematical, theoretical, and computational, and I have long been a robotics fanatic. Lately, I have been very curious about intelligence, learning, and reasoning, in both natural and artificial systems. Thus, I have been furthering my interest in cybernetics, and my research is mostly focused on NeuroAI; vis-à-vis motivating methods from observations in neuroscience to develop self-learning and self-exploring artificial intelligence (AI) for robotics applications.
My long-term goal is to develop autonomous systems that can generalize knowledge, transfer skills across tasks, exhibit flexible decision-making, and perhaps even show complex cognitive behaviors akin to biological intelligence.
Vice-versa, I am also interested in using these models to learn more about the processes in the brain. I have a secondary goal to study the emergence of impaired decision-making vis-à-vis neurocognitive disorders with experiments in robots. Using unified theoretical embodied models of perception and action-planning/decision-making, I’d like to study the potential causes of such impairments, with potential applications to improve cognitive behavioral therapies.
My Education and Career
I completed my Bachelor’s in Technology (B.Tech.) in 2018 from the Indian Institute of Technology (IIT) Roorkee in India, in Mechanical Engineering with a minor in Computer Science. I specialized in robotics and extensively indulged in competitive robotics for three years. I also did an industrial internship as an Automation Engineer at ITC Ltd. (an FMCG conglomerate in India). My Bachelor’s thesis was in Biorobotics, titled “Mathematical Modeling of Humanoid Robot Gait on a Vibrating Beam”.
After graduation, I transitioned to the industry and moved to Japan to work as a software engineer in Tokyo for three years in IIoT and AI, at JIG-SAW Inc. (a mid-size IT company) and briefly (part-time) at Resonest Corporation (a startup). In that capacity, I took varied roles spanning Data Analyst, Machine Learning Specialist, Embedded Systems Engineer, DevOps Engineer, Backend Web Developer, and Cloud Architect. I also worked as the lead technical advisor for JIG-SAW US for business expansion in the USA for a year and helped form partnerships with companies such as SAP, Tridium, and Datadog, among others.
During this time, I developed an interest in theoretical neuroscience and mathematical modeling of brain processes, and decided to go back to research in 2021, to further explore these interests. Subsequently, I did my Master’s in Neural Information Processing at the University of Tübingen in Germany, where I studied a range of subjects, from theoretical and computational neuroscience to brain computer interfaces and advanced machine learning.
In tandem with my studies, I worked as a computational research assistant at cellular neuroscience and brain imaging labs, with Prof. Dr. Andrea Burgalossi and Prof. Dr. Zhaoping Li, at the Center for Integrative Neuroscience (CIN) and the Max Planck Institute for Biological Cybernetics in Tübingen, respectively.
I also did an internship at a rehabilitation institute in Vitznau, Switzerland, in data analysis of non-invasive brain-computer interfaces (fNIRS), and developed fnirsPy, in collaboration with RELab (ETH Zürich) and the University of Zürich. (To see a list of my other open-source software projects for research, click here or visit my GitHub account.)
My Master’s thesis was at the intersection of model-based planning and reinforcement learning, inspired by cognitive science, titled “Building Visual Semantic Bias in Curious Exploration during Free Play”, which I completed in 2024 at the Autonomous Learning Lab at the Max Planck Institute for Intelligent Systems in Tübingen.
After graduation, from May–Dec 2024, I worked with Prof. Jun Tani in the Cognitive Neurorobotics Unit at Okinawa Institute of Science and Technology in Japan, on computational modeling of habitual and goal-directed behavior in obsessive-compulsive disorder (OCD), building upon a theory I motivated in an earlier essay on modeling OCD in self-exploring model-based reinforcement learning agents using methods from active inference and the free energy principle.
I am currently an ELLIS PhD student, working on developing human-like generalization characteristics for object manipulation in robots, with inspirations from cognitive-neuroscience. Specifically, I am investigating how robots can build object-centric abstract representations of the world through model-based methods in reinforcement learning and curiosity-driven self-exploration. I seek to create agents capable of self-learning and forming hierarchical abstractions, enabling efficient action planning, reasoning, and adaptation in complex environments.