Sungguk Cha
Biography
I am Researcher at the Service AI Research Team of LG Uplus
closely working with Dr. Mintae Kim and DongWook Kim.
Before that, I have spent wonderful times as a Technical Research Personnel for the military duty
working with
Dr. Cheoljong
Yang
and Jusung Lee
at NCSOFT,
and a number of fantastic colleagues at MINDs Lab Inc.
I obtained my M.S. degree in Ulsan National Institute of Science and Technology.
My research interests cover the broad area of machine learning, multi-modal learning, computer vision,
language modeling and quantitative finance.
My current position focuses on synthesizing train/evaluate data for (vision-)language models.
Publications
-
ReinPool: Reinforcement Learning Pooling
Multi-Vector Embeddings for Retrieval System
Sungguk Cha, DongWook Kim, Mintae Kim, Youngsub Han, Byoung-Ki Jeon, Sangyeob Lee
Preprint, 2026.
-
Annotation-Free Reinforcement Learning
Query Rewriting via Verifiable Search Reward
Sungguk Cha, DongWook Kim, Taeseung Hahn, Mintae Kim, Youngsub Han, Byoung-Ki Jeon
Preprint, 2025.
-
Visually Dehallucinative Instruction
Generation: Know What You Don't Know
Sungguk Cha, Jusung Lee, Younghyun Lee, Cheoljong Yang.
Preprint, 2024.
-
Visually Dehallucinative Instruction
Generation
Sungguk Cha, Jusung Lee, Younghyun Lee, Cheoljong Yang.
Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP),
2024.
-
Question Answer Instruction: Unlocking
Multi-Modal Large Language Model to Domain-Specific Visual Multi-Tasks
Jusung Lee, Sungguk Cha, Younghyun Lee, Cheoljong Yang.
Preprint, 2024.
-
Zero-Shot Semantic Segmentation via
Spatial and Multi-scale aware Visual Class Embedding
Sungguk Cha*, Yooseung Wang*.
Pattern Recognition Letters (PRL), Volume 158, June 2022, Pages 87-93.
Acknowledgements
-
Top Finisher in StarCarft AI Tournaments
Entry name: Zia Bot
Experiences
LG Uplus
August 2024 - Present
-
Service AI Research Team
(Jan. 2026 - Present)
-
Reference: Dr. Mintae Kim
-
Multimodal Embedding alignment and Interleaved Data Processing.
-
GUI Agent.
-
RAG AI Technical Team
(Jan. 2025 - Dec. 2025)
-
Reference: Dr. Mintae Kim
-
Multimodal RAG, document question answering
-
Multimodal RAG framework implementation including document embedding and question
answering.
-
Multi-turn RAG agent developments and evaluation dealing with the company secretary
documents. (reference Geun Lim)
-
Bulit the train/evaluate data for the multiturn query rewrite agent.
-
Built the evaluate data and the llm-as-judge model for the multiturn RAG scenario.
-
Closely worked with the related departments preparing benchmarks, data and handling
issues related to the RAG system.
-
Document Parsing (reference Soyeon Jo, Heonjin Ha)
-
Advancing document parsing (Document Layout Analysis(DLA), VLM parsing) : analysis,
evaluation and synthesizing.
-
Query Rewriting (reference DongWook
Kim, Taeseung Hahn)
(Data, Index)-agnostic query rewriting approach for retrieval system.
-
LLM Technical Team
(Aug. 2024 - Dec. 2024)
- Developed multi-turn RAG data synthesis approaches.
- MWC 2025 preparation. LLM for digital humanoid implementation.
OAQ Co. Ltd.
June 2024 - March 2025
NCSOFT
Sep. 2022 - May 2024
-
Technical Research Personnel for Alternative Military Service
(Sep. 2022 - Feb. 2024)
-
Image Generation Team, Multimodal AI Lab.
(Feb. 2024 - May 2024)
- VQA (VLM) research and development for digital humans.
- Multi-modal model research and development for game-domain. It includes multi-modal crawling
webscale data.
- Scale-up Data Curation for Generative Foundation Models: Replaced human-annotation in
captioning (tagging) and resulted in dramatic improvements in developing image generative
models.
-
Recognition Team, Vision AI Lab., NC Research
(Sep. 2022 - Feb. 2024)
- Multimodal Recognition AI: Generated multi-modal data for novel tasks. Researched on
evaluating large multimodal models. Responsible and dehallucinative multimodal data
generation.
- Conversational AI
- Facial Expression Recognition
MINDs Lab., Inc.
Sep. 2020 - Sep. 2022
- The company now goes by the name maum.ai.
-
Technical Research Personnel for Alternative Military Service
(Feb. 2021 - Sep. 2022)
-
Senior AI Scientist
(Mar. 2021 - Sep. 2022)
-
Co-led the research team (HR, managing, etc.)
-
Developed and implemented fundamental visual recognition applications, focusing on classification
such as firearms and OCR, detection including mover objects and car damage, segmentation with a
specialization in fashion items, and person re-identification for tracking corona-infected
individuals.
-
Contributed to Conversational AI, specializing in human-object interaction through achievements in
Visual Question Answering, ATM abnormal behavior detection, and human-object
detection/classification.
Statisitical Artificial Intelligence Lab
Jul. 2018 - Feb. 2019
- http://sailab.kaist.ac.kr/
- Adviser: Prof. Jaesik Choi
- Undergraduate Researcher
-
Research Topic:
Drivable Area Semantic Segmentation
Agents and Robotic Transportation Lab
Dec. 2017 - Feb. 2018
- Adviser: Prof. Tsz-Chiu Au
- Undergraduate Researcher
-
Research Topic:
StarCraft AI Algorithm
Machine Learning & Artificial Intelligence
Sep. 2017 - Dec 2017
SKT T-Brain
Mar. 2017
- Research service invited by Hyunsoo Kim (Superb AI CEO)
- Passed interviews
-
Research Topic:
Machine Learning Application for StarCraft
Professional Services
Conference Talks:
- LG Software Developer Conference (LG SDC) 2025.
Conference Reviewer:
- Association for the Advancement of Artificial Intelligence (AAAI) 2026.
- IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2025, 2026.
- IEEE International Joint Conference on Neural Networks (IJCNN) 2025, 2026.
Educations
Ulsan National Institute of Science and Technology
Mar. 2019 - Feb. 2021
-
Master of Science in Computer Science and Engineering
(GPA 3.94/4.30)
- Thesis: Zero-Shot Semantic Segmentation via Visual Class Embedding
- Full Merit based Scholarship
Ulsan National Institute of Science and Technology
Mar. 2016 - Feb. 2019
Mar. 2014 - Feb. 2015
- Bachelor of Science in
- Computer Science and Engineering (GPA 3.74/4.30)
- Electrical Engineering
- Cum Laude
- Full Merit based Scholarship
Honors & Awards
-
Dacon AI Bit Trader Competition 29th (Top 21%)
Mar 2021
-
NAVER LABS Mapping & Localization Challenge
8th place
Jun 2020
- Kaggle/iMaterialist (Fashion) 2019 at FGVC6/fashion item panoptic segmentation/(29th/194 (Top 15%))
- Kaggle/Google Analytics Customer Revenue Prediction/(570th/3611 (Top16%))
-
Student StarCraft AI Tournament
4th
Feb 2018
-
Naver UNIST Undergraduate Poster Award, Bronze Prize
Dec 2017
Topic: Reinforcement Learning in StarCraft in Real-Time using Deep Learning
-
Samsung Dream Class Mentoring Scholarship
Spring 2017
-
AAAI AIIDE StarCraft AI Competition, 28th
2017
-
IEEE CIG StarCraft AI Competition 3rd
2017
-
Student StarCraft AI Tournament 3rd
Feb 2017
-
AAAI AIIDE StarCraft AI Competition, 11th
2016
-
IEEE CIG StarCraft AI Competition 10th
2016
-
Korean Olympiad of Informatics Encouragement Prize
2013
Regional representative of Daejeon
-
Daejeon Olympiad of Informatics Gold Prize
2013
-
Korean Olympiad of Informatics Encouragement Prize
2012
Regional representative of Daejeon
-
Daejeon Olympiad of Informatics Gold Prize
2012
-
Korean Olympiad of Informatics Encouragement Prize
2010
Regional representative of Daejeon
-
Daejeon Olympiad of Informatics Gold Prize
2010
Teaching
-
[TA] UNIST/CSE241: Object Oriented Programming
Spring 2020
-
[TA] UNIST/CSE251: System Programming
Fall 2019
-
[Tutor] PyCon Korea
Aug 2019
Topic: Explainable AI
-
[TA] UNIST/ITP117: Introduction to AI Programming II
Spring 2019
-
[TA] UNIST/AHS161: What is "I"
Spring, Fall 2017
Skills
-
Languages
- Korean | English | Python | C/C++ | MFC
-
Frameworks
- Deep learning: PyTorch, Multimodal, RLHF/RLAIF, RAG agents
- Collaboration: ATLASSIAN, Tigris
- Developments: Data Curation, Interleaved Data Processing, Git, Docker, gRPC