№ 1986@jin-park
Jin Park
post-phd, hopping out of the lab. building a robot foundation-model startup that's mostly waiting for ai chips to land in seoul.
currently building
a manipulation policy foundation model trained on a few million teleoperated demonstrations. open weights for the small model, hosted for the large one.
asking for
anyone training large robotics models on >100 gpus. and: people who've thought carefully about teleoperation cost curves.
offering
i'll trade compute knowledge for application knowledge. specifically: what robotic tasks actually have product-market fit in 2026.
shipped — on file
- ·01PhD in robot learning, KAIST, 2024
- ·02co-founded a robot-learning startup (seed, 2025)