Derek Daehyun Ji
Vice President of Technology (Master), Samsung Electronics, Suwon, South Korea
Head of the Manufacturing & Material Handling AI Lab and Vertical AI 2 Team, leading AI and robotics technologies for production optimization and logistics automation.
Research areas: autonomous machines, robot manipulation, computer vision, embodied AI, autonomous driving, and robust learning.
captainzone [at] gmail.com · derek.ji [at] samsung.com
Scholar Snapshot
This section is manually synchronized from the public Google Scholar profile and can be updated whenever the profile numbers change.
About
Daehyun Ji is a technology leader and researcher working at the intersection of AI, computer vision, robotics, and real-world autonomous systems. He currently leads industrial AI initiatives at Samsung Electronics focused on production intelligence, logistics automation, and autonomous machines. Previously, he led AI technology development for ADAS and autonomous driving at Samsung Advanced Institute of Technology (SAIT), and also worked as a Visiting Researcher at Mila under Professor Simon Lacoste-Julien.
His broader mission is to create practical, high-impact AI that benefits people widely, with recent work spanning robot manipulation, vision-language-action systems, safety-aware imitation learning, navigation, 3D scene understanding, and domain adaptation.
Selected Recent Research
A recent embodied AI / robot manipulation work that integrates VLM reasoning with diffusion-based action models to coordinate arm movement and gripper control in dynamic environments.
Introduces a diffusion-transformer approach that combines wavelet-spectrum information for higher quality image super-resolution.
A safety-aware imitation learning framework using constrained diffusion policies for robot behavior under explicit safety boundaries.
Addresses online adaptation for VLN agents in unfamiliar environments using feedback-driven reinforcement learning at deployment time.
Proposes ProtoOcc, improving 3D occupancy prediction under computational constraints while preserving strong scene understanding performance.
Presents a practical framework for robust multi-view 3D object detection across unseen domains with label-efficient adaptation.
Focuses on online vectorized HD map construction for autonomous driving, addressing occlusion and temporal propagation challenges.
Introduces image-assisted unsupervised domain adaptation for LiDAR-based 3D object detection through cross-modal semantic transfer.
A cross-modal distillation framework that transfers structural and temporal cues from LiDAR teachers to image-based 3D detectors.
Selected Most-Cited Papers & Patents
Career Timeline
- Vertical AI 2 Team / Manufacturing & Material Handling AI Lab, Samsung Electronics March 2025 – Present · Vice President of Technology (Master), Head of Lab / Team Leads AI and robotics technologies for manufacturing, logistics, material handling, and device intelligence.
- Computer Vision TU, AI Research Center, SAIT November 2023 – 2024 · Vice President of Technology (Master) Led autonomous driving research and broader AI-based autonomous machine initiatives.
- Computer Vision TU, AI Research Center, SAIT March 2020 – November 2023 · Principal Researcher (Research Master) Project lead for autonomous driving with focus on perception, 3D scene understanding, and deployment-oriented AI.
- Mila – Quebec AI Institute January 2019 – January 2020 · Visiting Researcher Worked on noisy labels, uncertainty-aware learning, and knowledge distillation.
- A-Project Team, System LSI, Samsung Electronics September 2018 – May 2020 · Principal Engineer Developed ADAS SoC perception solutions.
- SW Solution Lab, Device & System Center, SAIT March 2015 – August 2018 · Principal Researcher Led deep learning and computer vision efforts for autonomous driving, including detection, segmentation, mapping, planning, and control.
- Mobile Communication Division, Samsung Electronics September 2009 – August 2015 · Senior SW Engineer Worked on Samsung Galaxy system software spanning kernel, device drivers, framework, and applications.
Education
- POSTECH
Integrated M.S./Ph.D. in Electrical Engineering
Major: Robotics & Control
Advisor: Prof. Sangchul Won - Kyungpook National University
B.S. in Electrical Engineering
Scholar Notes
- Total citations currently shown on the public profile: 2,019
- Profile: Samsung Electronics / AI Center / VP of Technology
- Main themes surfaced on Scholar: autonomous driving, robotics, deep learning, control theory
- Manual update tip: replace the numbers inside the citation badges when the profile changes
Honors & Awards
- DS Vision Award (Inclusion), Samsung Electronics, 2024
- SAIT Vision Award, Samsung Electronics, 2023
- Distinguished Engineer Award, DS Division, Samsung Electronics, 2022
- Boundless Search for Breakthroughs Award, SAIT, 2022
- Gold Award, Samsung Best Paper, 2018
- Samsung Future Creator Award, 2018
- Bronze Award, Samsung Best Paper, 2017
- Best AP Award, SAIT, 2017
Patent Highlights
- 84 citations · Method arranging user interface objects in touch screen portable terminal and apparatus thereof
- 71 citations · Apparatus and method for generating training data to train neural network determining information associated with road included in image
- 66 citations · Method and apparatus of detecting object of interest
- Device and method with camera pose estimation · US Patent App. 18/464,654 (2024)
- Method and apparatus with object detector training · US Patent App. 18/451,287 (2024)
- Method and device with traffic light state determination · US Patent App. 18/349,393 (2024)
Community & Talks
- Reviewer: NeurIPS, CVPR, ECCV, ICCV
- KCCV 2024 · “AI Research at Samsung”
- Ewha AI Industry-Academia Forum 2024 · “Generative AI in Autonomous Driving”
- KSAE Ewha Seminar 2024 · “Computer Vision for Autonomous Driving”
- IPIU 2023 · “SAIT Autonomous Driving”