My Dummy Talk
Learning How to Show a Card
Kundan Kumar
Research Scientist and GenAI Data Scientist
Hi! I’m Kundan Kumar, a Ph.D. candidate and researcher focused on building intelligent, secure, and adaptable AI systems for next-generation cyber-physical infrastructure. My work bridges deep reinforcement learning (DRL), multi-agent systems, large language models (LLMs), safe and explainable AI, and computer vision, with real-world applications in smart grids, autonomous vehicles, and critical infrastructure.
My Ph.D. research centers on physics-informed and safety-critical DRL frameworks that embed domain knowledge, safety constraints, and uncertainty into the learning process—enabling agents to make robust and interpretable decisions in dynamic, complex environments. My research within DRL focuses on techniques such as transfer learning, uncertainty quantification, and adversarial resilience to improve generalization, safety, and reliability across diverse tasks and environments.
I also develop LLM-integrated simulation frameworks for robotics and autonomous systems, combining vision-based perception, trajectory planning, and natural language reasoning to support high-level control and human-AI collaboration.
Beyond research, I enjoy sharing my insights through educational content on Substack and YouTube. Outside of work, I love cooking and skating. 🛼
0.1 Other Research Interests
Computer Vision
Visual perception, object detection, semantic segmentation, and sensor fusion for autonomous systems.
Statistical ML
Uncertainty quantification, probabilistic modeling, and data-driven inference in dynamic environments.
Self-Driving Systems
Learning-based control, trajectory planning, vision-based perception, and sensor fusion in autonomous driving environments.
0.2 Explore My Work
0.2.1 Blog
No matching items
0.2.2 Talks
0.2.3 Publications
No matching items
0.2.4 Projects
No matching items
News Highlights
| [Jul 2025] | Selected for the Cohere Machine Learning Summer School, hosted by Cohere Labs. |
| [Mar 2025] | Our paper on Advanced Semi-Supervised Learning with Uncertainty Estimation for Phase Identification in Distribution Systems has been accepted to IEEE PES General Meeting 2025. |
| [Jan 2025] | Our paper on Transfer Learning Enhanced Deep Reinforcement Learning for Volt-Var Control in Smart Grids has been accepted to IEEE PES Grid Edge Technologies Conference & Exposition 2025. |