- Chanwoo Park
- Research Intern at NAVER Cloud, Hyperclova-Vision
- Currently, I am working on:
- Multimodal Pretraining
- Research Intern(2026.01. ~ Current)
- At NAVER Cloud
- Multimodal Pretraining: Optimizing VLM training workload for Hyperscale GPU systems.
- Research Intern(2025.11. ~ 2025.12.)
- At SK Telecom
- Building data for pre-training, post-training LLMs.
- Researcher (2022.11. ~ 2025.10.)
- At Thunder Research Group
- Improving model quality with corpus engineering
- Most Used Languages:
C++,Python - Advisor: Prof. Jaejin Lee
- TmaxTibero (2022.07. ~ 2022.08.)
- Summer Internship at DevOps Team
- Composing Network of DBMS containers and Parallelizing test tasks for Test Automation
- Most Used Technology:
PythonParallel Programming,Shell,Docker,Docker Compose
- Undergraduate Research Opportunities Program (2022.03. ~ 2022.06.)
- At CSAP Lab
- Implemented DNN operation with cuda programming apis
- Evaluated DNN Post-training Quantization results
- Most Used Languages:
C++,Python - Advisor: Prof. Bernhard Egger
- Beyond Line-Level Filtering for the Pretraining Corpora of LLMs
- Chanwoo Park, Suyoung Park, Yelim Ahn, Jongmin Kim, Jongyeon Park, Jaejin Lee
- On arXiv: https://arxiv.org/abs/2510.24139
- Ko-MuSR: A Multistep Soft Reasoning Benchmark for LLMs Capable of Understanding Korean
- Chanwoo Park, Suyoung Park, JiA Kang, Jongyeon Park, Sangho Kim, Hyunji M. Park, Sumin Bae, Mingyu Kang, Jaejin Lee
- On arXiv: https://arxiv.org/abs/2510.24150
- Enhancing Face Recognition Accuracy for Under-Display Cameras via Image Restoragtion
- Kyusu Ahn, Chanwoo Park, Jaejin Lee
- (Distinguished Paper)On Society for Information Display's Display Week 2025
- UDC-VIT: A Real-World Video Dataset for Under-Display Cameras
- Kyusu Ahn, Jisoo Kim, Hyungyu Lee, Byeonghyeon Ko, Chanwoo Park, Jaejin Lee
- To Appear on ICCV 2025 : On arXiv: https://arxiv.org/abs/2501.18545
- Integrating Spatial and Frequency Information for Under-Display Camera Image Restoration
- Kyusu Ahn, Jinpyo Kim, Chanwoo Park, Jisoo Kim, Jaejin Lee
- On arXiv: https://arxiv.org/abs/2501.18517
- Analysis of Large Language Model Performance According to Korean-English Training Data Ratio in Continual Learning
- Jinpyo Kim, Gyeongje Cho, Chanwoo Park, Jongwon Park, Jaejin Lee
- On Korea Software Congress 2024
- Exploring the Impact of Pre-training Corpora on the Performance of Language Models
- Yujin Jeong, Jaehwan Lee, Chanwoo Park, Donghun Choi, Jaejin Lee
- On Korea Computer Congress 2024
- UDC-SIT: A Real-World Dataset for Under-Display Cameras
- Kyusu Ahn, Byeonghyun Ko, Hyungyu Lee, Chanwoo Park, and Jaejin Lee
- On the 37th Conference on Neural Information Processing Systems, Datasets and Benchmarks
- Applying Influence Range to Under-Display Camera Image Restoration
- Kyusu Ahn, Byeonghyun Ko, Hyungyu Lee, Chanwoo Park, and Jaejin Lee
- On the 23rd International Meeting on Information Display
- Bacchus (2021.03. ~ 2023.12.)
- Was in charge of onboarding members and member relations
- Managed more than 60 machines in the department.
- Most Used Technology:
ansible
- Seoul National University (2023.09. ~ 2025.08.)
- Master's course in Computer Science and Engineering
- Advisor: Prof. Jaejin Lee
- Thesis: Towards Higher Generative Question-Answering Accuracy: Punctuation Mark Pattern-based Line Filtering on Korean Web Text
- Seoul National University (2017.03. ~ 2023.08.)
- B.A. in Business Administration and Computer Science and Engineering
- Best Thesis Paper in College of Business Administration
- Thesis: Analysis of IPO Underpricing and After-Market Price Fluctuation
- CSE Thesis
- Thesis: Applying activation recomputation for CNN training
- Summa Cum Laude
- Hankuk Academy of Foreign Studies (2014.03. ~ 2017.02.)
- Humanities and Social Science Course
- Foundation of Computing, Tutor (2021.09. ~ 2023.06.)
- Teaching language:
Python - Practice session teaching assistant
- Teaching language:
- Teaching Assistance for CSE Courses
- Algorithms (2022.09 ~ 2022.12)
- System Programming (2025.03 ~ 2025.06)
- Server Manager at Thunder Research Group
- Built/Managed Peta Scale(1.2PB) Distributed File System(Lustre)
- Managed a GPU Cluster with more than 63 GPU Nodes, 278 GPUs, and 300+ Users



