Research internship
RideFlux · VVSW & Perception Unit
Worked on ML-based Planning, Scenario Tagging, and MLOps.

I am a Ph.D. student working with Prof. Soonmin Hwang on representation learning for autonomous driving. I started my doctoral studies in September 2026, after receiving my B.S. in Computer Engineering from Dongguk University and my M.S. in Automotive Engineering from Hanyang University. Previously, I interned with the VVSW & Perception Unit at RideFlux, working on ML-based planning, scenario tagging, and MLOps. During my undergraduate studies, I conducted medical imaging research. My research interests include End-to-End Autonomous Driving, Closed-loop Evaluation, 3D Gaussian Splatting, and Planning-aligned Pretraining.
PAVER is now available on arXiv.
RSD-BEV is accepted to ICPR 2026.
Self-Guided Low Light Object Detection Framework is accepted to ICLR 2026.
Our team won 3rd place in the Waymo Vision-based End-to-End Driving Challenge.
RideFlux · VVSW & Perception Unit
Worked on ML-based Planning, Scenario Tagging, and MLOps.
IRCV Lab · Hanyang University
End-to-end autonomous driving, BEV perception, and 3D Gaussian Splatting. Advised by Prof. Soonmin Hwang.
Machine Learning Lab · Dongguk University
3D medical image segmentation. Advised by Prof. Jihie Kim.
Waymo Open Dataset Challenge
Our team placed third in the vision-based end-to-end driving track of the Waymo Open Dataset Challenge at the CVPR 2025 Workshop.
KSAE Autumn Conference
Received the Best Poster Award for our research presentation at the 2024 KSAE Autumn Conference.
ICIP & Capstone Project
Won first prize for Virtual Docent, a project combining language, speech, and facial animation to let artists narrate their work.
International Collegiate EV Autonomous Driving Contest
Our MACARON 4.0 team won first prize in the inaugural International Collegiate EV Autonomous Driving Contest.
Advisor: Soonmin Hwang
Advisor: Soonmin Hwang
Developing pretraining methods for knowledge-distillation-based end-to-end driving.
Led development of pseudo-LiDAR data pipelines, BEV planning and reward-model experiments.
Developed VAD-based multi-agent prediction and built an OpenScene data pipeline for longer driving sequences.
Built a TensorRT/C++ pipeline for LiDAR-based 3D detection, tracking and collision-risk assessment.
Contributed to vision-based end-to-end driving, evaluation and pseudo labeling for the 3rd-place team.
Worked on pseudo labeling and model quantization for on-device forward-collision prevention.
Developed distance-aware label smoothing and training pipelines for diffusion-based multi-organ CT segmentation.
Led GPS-free driving development using visual localization, stereo-depth obstacle avoidance and real-time detection.
Implemented camera/LiDAR parking and delivery missions, lane detection and vehicle communication.
English · Advanced Low
English · 775 / 990
English · Intermediate High