Sungho Park
I am a Ph.D. student in the Data Systems Lab at POSTECH, advised by Prof. Wook-Shin Han. My research centers on developing self-improving AI agents, inspired by Sun Tzu’s The Art of War: "If you know the enemy and know yourself, you need not fear the result of a hundred battles." I view the path to Artificial General Intelligence (AGI) as rooted in a model's ability to autonomously identify its weaknesses—"knowing oneself"—and iteratively enhance its performance.
As a primary testbed for this self-improving framework, I am currently focusing on the development of Agentic RAG systems for deep reasoning across large-scale multi-modal data, such as complex tables and unstructured text. This work connects multi-hop retrieval with broad information synthesis, empowering agents to traverse intricate data environments with exceptional accuracy.
News
SPARTA: Scalable and Principled Benchmark of Tree-Structured Multi-hop QA over Text and Tables
A paper on tree-structured multi-hop QA has been accepted to the International Conference on Learning Representations (ICLR) 2026.
Microsoft Internship
Received an internship offer from Microsoft. I will be joining the Data, Knowledge, and Intelligence (DKI) team in Beijing, China from March to September 2026.
BK21 Best Paper Award (Grand Prize)
Received the BK21 Best Paper Award (Grand Prize) from POSTECH Graduate School of AI for HELIOS: Harmonizing Early Fusion, Late Fusion, and LLM Reasoning for Multi-Granular Table-Text Retrieval.
Past notices
A paper on table-text retrieval has been accepted to the Association for Computational Linguistics (ACL) 2025.
KDD Cup Meta CRAG 2024: Three-step Question-Answering Framework
Our team dRAGonRAnGers won 🏆 1st place in Comparison & Post-processing Tasks at KDD Cup 2024.
Oracle Labs Internship
Received an internship offer from Oracle Labs (Switzerland) for Summer 2023 on LLM Research.
![[width=1500]](/_astro/8b1a613d-425e-4fb6-a8fd-040c7360e3f8.BXiirjyx_M6Fqd.webp)