Junsung Park
Junsung Park is an undergraduate researcher at Seoul National University's Robot Learning Lab, where he works under Professor Songhwai Oh on reinforcement learning and learning-based control for robot autonomy. He is also affiliated with UC San Diego as a research assistant and has prior research experience at Seoul National University's Soft Robotics and Bionics Lab and Robotics Laboratory. His work focuses on building generalizable robot autonomy for robust interaction in dynamic environments, with interests spanning humanoid robots, wearable interfaces, and offline inverse reinforcement learning.
“Simile is building a foundation model of human behavior that can be used to create simulations of individuals, subpopulations, and entire ecosystems. They want models that are biased and fallible in the same ways humans are.”
Source→“My fundamental thesis here is for AI companies of this generation, you need to have an interesting data strategy that's going to be defensible... Do you have access to data that no one else has access to? Do you know how to collect data that is very hard to collect?”
Source→“I think there is a world in which in about two, three years, we're running a single simulation session. And that's going to take 10, $20 million to run a single session. But it's going to be so valuable that people will pay $100 million for it.”
Source→“Edged, a chip/hardware AI company that recently came out of stealth, was identified by Jun as one of the most exciting teams in the chip/inference layer space.”
Source→“Jun Sung Park is Founder and CEO of Simile. Former Stanford PhD researcher and lead of the Smallville generative agent simulation experiment (2023). Michael Bernstein, Percy Liang, and Lainey Allen are co-founders.”
Source→“Co-lead author on TactX. Dual affiliation suggests cross-institutional collaboration between US and Korean robotics ecosystems — relevant for investors tracking global Physical AI talent flows”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.