Skip to content

Yunsung Lee

AI/ML Research Engineer

Contact Information

Professional Summary

AI/ML research engineer with strong research foundations (10+ publications in top-tier conferences, 1.5K+ citations) and practical experience deploying commercial AI products (over 5M MAUs). Currently leading research initiatives at Maum.ai's WoRV team, specializing in Embodied AI, autonomous agents, and robotics foundation models. Expertise spans multimodal AI, autonomous systems, and translating cutting-edge research into real-world applications.

Core Competencies

Research Areas: Vision-Language Multimodal AI, Diffusion Models, Autonomous Agents, Robotics Foundation Models

Technical Skills: Python, PyTorch, Git, Docker, FastAPI, Elasticsearch, ML Model Deployment

Leadership: Research Team Management, Cross-functional Collaboration, Technical Mentoring

Professional Experience

Alternative Military Service

Positions at Wrtn Technologies and Riiid serve as Alternative Military Service (Technical Research Personnel, 전문연구요원) until April 14, 2025, combining mandatory service with R&D in strategic industries.

Head of Research, WoRV Team MaumAI · Seongnam, South Korea · May 2025 - Present

Leading research initiatives for WoRV (World Model for Robotics and Vehicle Control), Maum.ai's flagship Embodied AI research organization. Overseeing development of foundation models that integrate language, vision, and action for robotics and autonomous driving applications. Driving projects including SketchDrive navigation agents, autonomous agricultural machinery, and open-world agents research.

AI Engineer Wrtn Technologies · Seoul, South Korea · Oct. 2023 - Apr. 2025

Artificial intelligence engineer specializing in memory and personalization for autonomous agents, and multimodal capabilities for wrtn. Led core development efforts, spanning ML technology research and implementation to backend engineering. Drove key functionalities of the project with focus on autonomous agent capabilities.

Research Scientist Riiid · Seoul, South Korea · Apr. 2022 - Oct. 2023

Computer vision research scientist focusing on educational AI applications. Math problem image retrieval for AI:R Math. English vocabulary visualization with text-to-image diffusion models (Santa).

ML Research Scientist Scatter Lab · Seoul, South Korea · Jul. 2021 - Mar. 2022

Researched vision-and-language multimodal dialogue system for the chatbot "Luda Lee". Reference: Make Luda see, Naver Deview 2023.

OCR Team Intern CLOVA, NAVER Corp · Seongnam, South Korea · Sep. 2020 - Mar. 2021

Researched self-supervised representation learning, domain generalization, and data augmentations for document images.

ML Team Intern HYPERCONNECT · Seoul, South Korea · Jul. 2020 - Aug. 2020

Researched adversarial robust semi-supervised learning.

Intern, Web Developer Algorithm Labs · Seoul, South Korea · Dec. 2016 - Feb. 2017

On-site internship with focus on competitive programming and web development. Wrote competitive programming problems and contributed to web front-end development.

Education

M.Sc. Computer Science Korea University · Seoul, South Korea · Mar. 2019 - Feb. 2022 Advisor: Seungryong Kim, Jaegul Choo

Visiting Scholar Carnegie Mellon University · PA, USA · Jan. 2020 - Jul. 2020 Artificial Intelligence, Language Technologies Institute. Sponsored by IITP under South Korean Government.

B.Sc. Computer Science Hanyang University · Seoul, South Korea · Mar. 2014 - Feb. 2019

Publications

Research Impact

My publications have accumulated over 1,500 citations on Google Scholar, demonstrating significant influence in machine learning and computer vision research.

!!! note "Publication Notes" * denotes equal contribution (co-first authors). ^† denotes co-corresponding authorship.

2025

  • Yohan Lee*, Sungho Park*, Sangwoo Han*, Yunsung Lee*^†, Yongwoo Song, Adam Lee, Jiwung Hyun, Jaemin Kim, Seungtaek Choi, HyeJin Gong^† "SAFARI: Sample-specific Assessment Framework for AI in Real-world Interactions," Findings of Annual Conference of the North American Chapter of the Association for Computational Linguistics (Findings of NAACL'25), 2025

  • Jin-Young Kim*, Soonwoo Kwon*, Hyojun Go*, Yunsung Lee, and Seungtaek Choi, "ScoreCL: Augmentation-Adaptive Contrastive Learning via Score-Matching Function," Machine Learning (Springer Journal), 2025

2024

  • Yunsung Lee*, Jin-Young Kim*, Hyojun Go*, Myeongho Jeong, Shinhyeok Oh, and Seungtaek Choi "Multi-Architecture Multi-Expert Diffusion Models," Annual AAAI Conference on Artificial Intelligence (AAAI'24), 2024

2023

  • Hyojun Go*, Jinyoung Kim*, Yunsung Lee*, Seunghyun Lee, Shinhyeok Oh, Hyeongdon Moon, and Seungtaek Choi "Addressing Negative Transfer in Diffusion Models," Conference on Neural Information Processing Systems (NeurIPS'23), 2023

  • Shinhyeok Oh*, Hyojun Go*, Hyeongdon Moon, Yunsung Lee, Myeongho Jeong, Hyun Seung Lee, and Seungtaek Choi, "Evaluation of Question Generation Needs More References," Findings of Annual Meeting of the Association for Computational Linguistics (Findings of ACL'23), 2023

  • Hyun Seung Lee*, Seungtaek Choi*, Yunsung Lee, Hyeongdon Moon, Shinhyeok Oh, Myeongho Jeong, Hyojun Go, and Christian Wallraven "Cross Encoding As Augmentation: Towards Effective Educational Text Classification," Findings of Annual Meeting of the Association for Computational Linguistics (Findings of ACL'23), 2023

  • Hyojun Go*, Yunsung Lee*, Jin-Young Kim*, Seunghyun Lee, Myeongho Jeong, Hyun Seung Lee, and Seungtaek Choi "Towards Practical Plug-and-Play Diffusion Models," The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2023 (CVPR'23), 2023

2022

  • Yunsung Lee*, Gyuseong Lee*, Kwangrok Ryoo*, Hyojun Go*, Jihye Park*, Seungryong Kim "Towards Flexible Inductive Bias via Progressive Reparameterization Scheduling," ECCV 2022 Visual Inductive Priors for Data-Efficient Deep Learning Workshop (ECCVW'22), 2022

2021

  • Seokju Cho*, Sunghwan Hong*, Sangryul Jeon, Yunsung Lee, Kwanghoon Sohn, and Seungryong Kim "CATs: Cost Aggregation Transformers for Visual Correspondence," Conference on Neural Information Processing Systems (NeurIPS'21), 2021

  • Junbum Cha, Hancheol Cho, Kyungjae Lee, Seunghyun Park, Yunsung Lee, and Sungrae Park. "SWAD: Domain Generalization by Seeking Flat Minima," Conference on Neural Information Processing Systems (NeurIPS'21), 2021

  • Yunsung Lee, Teakgyu Hong, Han-Cheol Cho, Junbum Cha, and Seungryong Kim. "HoughCL: Finding Better Positive Pairs in Dense Self-supervised Learning," ICML 2021 Workshop: Self-Supervised Learning for Reasoning and Perception (ICMLW'21), 2021

  • Yunsung Lee, Teakgyu Hong, and Seungryong Kim. "Data Augmentations for Document Images," The AAAI-21 Workshop on Scientific Document Understanding (AAAIW'21), 2021

2020

  • Seokeon Choi, Junhyun Lee, Yunsung Lee, and Alexander Hauptmann. "Robust Long-Term Object Tracking via Improved Discriminative Model Prediction," The ECCV-20 Workshop on Visual Object Tracking Challenge (ECCVW'20), 2020

  • Junsoo Lee*, Eungyeup Kim*, Yunsung Lee, Dongjun Kim, Jaehyuk Chang, and Jaegul Choo "Reference-Based Sketch Image Colorization using Augmented-Self Reference and Dense Semantic Correspondence," IEEE Conference on Computer Vision and Pattern Recognition (CVPR'20), 2020

2019

  • Sookyung Kim,* Sunghyun Park,* Sunghyo Chung,* Joonseok Lee, Yunsung Lee, Hyojin Kim, Mr Prabhat, and Jaegul Choo "Learning to Focus and Track Extreme Climate Events," British Machine Vision Conference (BMVC'19), 2019. Accepted as Spotlight Presentation (6.9% acceptance rate for spotlight papers), 2019.

Honors & Awards

Competition Awards

  • 5th place, RLT-DiMP team, VOT-Long-Term, ECCVW Visual Object Tracking Challenge, 2020
  • 4th place, HYU Programming Contest (Advanced Division), Seoul, Korea, 2015
  • Dean's citation for ACM-ICPC Daejeon Regional, Hanyang University, Seoul, Korea, 2014

Other Activities

Conference Reviewer

Top-tier ML/CV Conferences:

  • The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2023 - )
  • The IEEE/CVF International Conference on Computer Vision (ICCV) (2023 - )
  • Advances in Neural Information Processing Systems (NeurIPS) (2024 - )
  • European Conference on Computer Vision (ECCV) (2024 - )
  • The International Conference on Learning Representations (ICLR) (2024 - )
  • ACL Rolling Review for ACL, EMNLP (2024 - )

Selected Open-source Contribution

Open-World Agents

Main contribution member of open-world-agents org, open-world-agents, desktop-env. An open-source organization that aims to contribute useful tools for open-world agent R&D, including a real-time, high-frequency, real-world desktop environment suitable for desktop-based ML development (agents, world models, etc.).

Mem0

Contributor of mem0, the Memory layer for your AI apps.

Awesome Vision Transformers

Creator of awesome-visual-representation-learning-with-transformers, an awesome repository (with 200+ stars) from 2021, when Vision Transformer research was in its early renaissance.

Professional Organizations & Activities

PR12, Tensorflow Korea

Member | Jun. 2020 - Present

Active member of TensorFlow Korea, the country's largest machine learning research community, participating in an advanced study group focused on cutting-edge ML research.

TEDxHanyangU

Organizer | Sep. 2017 - Jun. 2018

  • Experience Catalyst, 2018
  • Web Engineer, 2017

SW Maestro

Trainee | Aug. 2017 - Jan. 2018

Talent training program by Ministry of Science and ICT Korea (under South Korean Government). Did a CPA document automation project. Learned basic computer vision and machine learning. Motivated to go to graduate school.

ALOHA (Algorithm research team)

Team Leader | Nov. 2015 - Oct. 2016

  • Taught algorithms to team members. Hosted several programming contests
  • Test writer & Presenter, The First KSH (Korea, Sookmyung W, Hanyang Univ.) Algorithm Camp
  • Chief test writer, HYU Programming Contest

Home Bartender

Licensed Craftsman Bartender

Licensed Craftsman Bartender certified by the South Korean government, combining precision and creativity in both professional and recreational pursuits.

Contact

For more information or collaboration opportunities, feel free to reach out through any of the contact methods listed above.