Shivank Gupta

I'm a research engineer at Penn Medicine and GRASP Lab, building autonomous surgical workflows and physical intelligence for the OR. I recently graduated from UPenn with a Master's in Robotics advised by Dr. Daniel Hashimoto and Kostas Daniilidis. I completed my undergrad at Worcester Polytechnic Institute (WPI) in Mechanical and Robotics Engineering.

Earlier work spans R&D Engineering at Boston Scientific, prototyping endoscopic medical devices and authoring FDA 510(k) test methods. I've also worked at Molex as an Engineering Intern working on automated hardware testing systems and manufacturing techniques for robotic cables.

These days, I enjoy spending my free time cooking for friends and family, finding new coffee spots, playing basketball, and doomscrolling through LinkedIn.

some active interests

autonomous systems, world foundation models, computer vision, surgical intelligence, mechanical design, physical AI

thesis

Physical AI can only be as strong as its multimodal reasoning. Humans spend a lifetime building embodied intelligence before acting on the world, and today's models are asked for that same reasoning without it. Real deployment is where you learn whether a system can actually think and act, or was only ever taught to perform.