Juwan Koh

- Teleoperated AI worker with leader-follower robot arms to collect manipulation data, gaining hands-on understanding of robot kinematics and motion constraints
- Built and developed a clustering-based outlier detection pipeline and camera-latency detection tool, generalized across multiple teleoperation datasets
- Trained GR00T models on curated datasets, tuned key hyperparameters, and refined data quality through image segmentation and labeling
- Achieved the team's first successful autonomous visuo-tactile dexterous manipulation task using TactileACT and ViTacFormer

• Selected as one of eight participants for the inaugural Engineering Career Education Program for Youth by ACEWORKS, one of the most prominent startups in Korea.
• Researched and presented domestic and international market trends, policies, and corporate strategies in the mobility industry aligned with climate change initiatives.
• Observed modeling and validation processes for software-defined vehicles (SDVs) and collaborated with team members to align virtual client requirements with ACEWORKS programs, developing tailored solutions.