I am a Senior Researcher at Samsung Research America (SRA), working on robotic manipulation, navigation, agentic memory, and multimodal intelligence for edge devices. Before joining SRA, I was a postdoctoral researcher at POSTECH, where I earned my Ph.D. in Computer Science and Engineering under Professor Minsu Cho. During my research journey, I collaborated with Google Research on multimodal video understanding with Cordelia Schmid, and with Microsoft Research Asia on bio-inspired machine vision with Chong Luo.
My research interests center on embodied intelligence, particularly vision-language-action models for robotic manipulation and navigation. I am interested in equipping such systems with short- and long-term agentic memory, together with efficient bio-inspired perception such as foveation, to enable practical on-device AI. Looking ahead, I envision future AI systems operating in hybrid edge-cloud settings, where lightweight on-device agents handle latency- and privacy-sensitive perception, while larger cloud-based agents provide deeper reasoning and broader world knowledge when needed. My goal is to build AI systems that can continuously perceive, remember, reason, and act in the physical world.
Work experience
-
Senior Researcher, Samsung Research America- Robotic manipulation (VLA) and navigation (VLN) with agentic memory
- Long-horizon streaming video-language understanding
- Bio-inspired visual reasoning for efficient vision-language models (VLMs) · Published FoveateR
- Multimodal retrieval for edge devices with Suren Kumar · Published GoldiCLIP
-
Postdoctoral Researcher, POSTECH- 3D geometric shape assembly, foundation models for robotics
- Collaborated with Professor Minsu Cho
- Published PMTR at ICML 2024 and Combinative Macthing at ICCV 2025
-
Student Researcher, Google Research- Modular reasoning for video question answering
- Collaborated with Shyamal Buch, Arsha Nagrani, and Cordelia Schmid
- Published MoReVQA at CVPR 2024
-
Research Intern, Microsoft Research Asia -
Associate Software Engineer, Electronic Arts Korea Implemented and maintained client-side UI components for FIFA Online 3.
Publications
- arXiv 2026 Project Page arXiv Bibtex GoldiCLIP: The Goldilocks Approach for Balancing Explicit Supervision for Language-Image Pretraining
- arXiv 2026 Project Page arXiv Bibtex Foveated Reasoning: Stateful, Action-based Visual Focusing for Vision-Language Models
- ICCV 2025 highlight Project Page arXiv Bibtex Combinative Matching for Geometric Shape Assembly
- ICML 2024 Project Page arXiv Poster Open Review Bibtex Code 3D Geometric Shape Assembly via Efficient Point Cloud Matching
- CVPR 2024 Project Page arXiv Poster Video Open Access Bibtex MoReVQA: Exploring Modular Reasoning Models for Video Question Answering
- WACV 2024 oral presentation best paper finalist Project Page arXiv Bibtex Code Efficient Semantic Matching with Hypercolumn Correlation
- TPAMI 2023 Project Page arXiv IEEE Xplore Bibtex Code Convolutional Hough Matching Networks for Robust and Efficient Visual Correspondence
- NeurIPS 2022 Project Page arXiv Poster OpenReview Bibtex Code Peripheral Vision Transformer
- CVPR 2022 Project Page arXiv Open Access Bibtex Code TransforMatcher: Match-to-Match Attention for Semantic Correspondence
- ICCV 2021 Project Page arXiv Video Open Access Bibtex Code Relational Embedding for Few-Shot Classification
- ICCV 2021 Project Page arXiv Poster Open Access Bibtex Code Hypercorrelation Squeeze for Few-Shot Segmentation
- CVPR 2021 oral presentation Project Page arXiv Poster Video Open Access Bibtex Code Convolutional Hough Matching Networks
- WACV 2021 Open Access Bibtex Learning to Distill Convolutional Features into Compact Local Descriptors
- ECCV 2020 Project Page arXiv Video ECVA Bibtex Code Learning to Compose Hypercolumns for Visual Correspondence
- arXiv preprint 2019 Project Page arXiv Bibtex Download Dataset SPair-71k: A Large-scale Benchmark for Semantic Correspondence
- ICCV 2019 Project Page arXiv Poster Open Access Bibtex Code Hyperpixel Flow: Semantic Correspondence with Multi-layer Neural Features
Education
-
-
The Pennsylvania State University, B.S. in Computer Science and Engineering
Sep 2011 – Dec 2014
Honors and Awards
- IPIU Best Paper Award (2024), 3D Geometric Shape Assembly via Efficient Point Cloud Matching
- CSE Outstanding Research Award (2024), in CSE at POSTECH
- Best Ph.D. Dissertation Award (2024), in CSE, GSAI, EE, CITE at POSTECH
- BK21 Outstanding Paper Awards (2023)
- Google PhD Fellowship (2022), Machine Perception, Speech Technology, and Computer Vision
- Outstanding Reviewer at CVPR (2022), awarded to top 5% reviewers
- POSTECHIAN Fellowship Award (2022)
- BK21 Outstanding Paper Awards (2022)
- Qualcomm Innovation Fellowship Korea (2021), Convolutional Hough Matching Networks
- Outstanding Reviewer at ICCV (2021), awarded to top 5% reviewers
- The 1st POSTECH Research Performance Contest (2021), fourth prize
- NAVER Ph.D. Fellowship Award (2020)
- Best Term Project Presentation Award (2019), in-class achievements
Professional services
- Webmaster, International Conference on Computer Vision (ICCV) 2019
- Regular reviewer: CVPR, ICCV, ECCV, NeurIPS, ICML, AAAI, TPAMI, IJCV, WACV, MVA
Military obligation
- Rifleman, 59 ASP & 102 Replacement Battalion, Chuncheon, Korea · Jun 2015 – Mar 2017
Language skills
- Korean (native)
- English (fluent)