Hi, I'm Guangxun Zhao,

About Me

Hi, I'm Guangxun Zhao, a PhD student in AI-based Convergence at Dankook University in Yongin, Korea.

I have been a student researcher in the System Software Laboratory since July 2021, advised by Prof. Seehwan Yoo and Prof. Jongmoo Choi.

My research focuses on LSM-tree key-value stores, learned indexes, and storage systems. I study how storage structures and machine learning can improve system performance.

You can find my code on GitHub. Outside research, I enjoy fitness, learning languages, and travel.

News

Dankook University News

VLDB 2026 research featured by Dankook University

Dankook University highlighted our paper, “Rethinking Learned Index and LSM-tree Integration,” in its coverage of the team's two VLDB 2026 papers on AI-driven database systems.

Read university news (Korean)

Research Interests & Skills

Systems for ML

Key-value stores and LSM-tree storage engines, including LevelDB and RocksDB.

AI for Systems

Learned indexes and reinforcement learning; piecewise linear regression (PLR) and PGM indexes.

Systems

Filesystems, Linux Kernel, Operating Systems, I/O Stacks.

Languages

Chinese (native), Korean (TOPIK Level 6, 258/300), English, and Japanese.

Education

  1. Sep 2024 – Present

    PhD in AI-based Convergence

    Dankook University

    Yongin, Korea

    Full Tuition Scholarship

  2. Mar 2023 – Aug 2024

    MS in AI-based Convergence

    Dankook University

    Yongin, Korea

    Full Tuition Scholarship

  3. Mar 2019 – Feb 2023

    BS in Software Science

    Dankook University

    Yongin, Korea

    50% Tuition Scholarship

Research & Teaching Experience

  1. Jul 2021 – Present

    Student Researcher

    System Software Laboratory, Dankook University

    Advisors: Prof. Seehwan Yoo and Prof. Jongmoo Choi

    • Key-value stores: SkipList, KV cache, compaction, SST files, and Bloom filters.
    • Linux kernel: CPU scheduler, memory allocator, and block I/O stack.
    • Learned and traditional index structures.

  2. Spring 2025

    Teaching Assistant

    Open Source SW Analysis (Bigdata), Dankook University

    • Guided Skip List and B+ Tree implementation, lab sessions, and code review.
    • Assisted with assignment and exam design, proctoring, and grading.

Publications

International Conferences

  1. [VLDB 2026] Rethinking Learned Index and LSM-tree Integration.
    Guangxun Zhao, Yongjie Zhu, Charles Jaranilla, Seehwan Yoo, Jongmoo Choi
    PDF GitHub Code
  2. [IEEE CCGrid 2026] ARC: Adaptive Resource Coordination for Write Stall Mitigation in LSM-Tree.
    Guangxun Zhao, Yongjie Zhu, Suhwan Shin, Seehwan Yoo, Jongmoo Choi
    Paper GitHub Code
  3. [IEEE BigComp 2024] ThanosKV: A Holistic Approach to Utilize NVM for LSM-tree based Key-Value Stores.
    Guangxun Zhao, Hojin Shin, Seehwan Yoo, Seongje Cho, Jongmoo Choi
    Paper
  4. [ACM RACS 2024] TodoroKV: Tiered Storage Distribution and Refinement for KV Separation in LSM-tree.
    Charles Jaranilla, Guangxun Zhao, Seehwan Yoo, Seongje Cho, Jongmoo Choi
    Paper

International Journals

  1. [Electronics (MDPI) 2025] A Framework for Integrating Log-Structured Merge-Trees and Key–Value Separation in Tiered Storage.
    Charles Jaranilla, Guangxun Zhao, Gunhee Choi, Sohyun Park, Jongmoo Choi
    PDF Paper

Domestic Conferences

  1. [KCC 2025] Rethinking SSTable Size: A Level-Sensitive Approach to LSM-tree Optimization.
    Guangxun Zhao, Yongjie Zhu, Yongmin Lee, Dayeon Wee, Yeongyu Choi, Jongmoo Choi
    Best Presentation Paper Award
  2. [KCC 2025] Predictive Models for KVS Performance Analysis and Tuning
    Charles Jaranilla, Guangxun Zhao, Yeongyu Choi, Yongjie Zhu, Jongmoo Choi
    Best Paper Award
  3. [KCC 2025] From Distance to Distribution: A Probabilistic Cluster Selection Method for High-Dimensional Vector Search
    Yeongyu Choi, Guangxun Zhao, Charles Jaranilla, Yongjie Zhu, Yongmin Lee, Jongmoo Choi
    Best Paper Award
  4. [KCC 2024] Quantitative Analysis of Learned Index in Bourbon with Larger SSTable Mechanism
    Guangxun Zhao, Hojin Shin, Seehwan Yoo, Jongmoo Choi
  5. [KCC 2024] Performance Analysis of Multi-Level Key-Value Separation in BlobDB
    SuGie Park, Guangxun Zhao, SoungHyoun Lee, JongMoo Choi
  6. [KCC 2024] Analysis of Performance according to Blob File Size of BlobDB based on RocksDB
    Min Seong Kim, Guangxun Zhao, Gunhee Choi, Jongmoo Choi, SeeHwan Yoo
  7. [KCC 2024] Balance-SkipList: Balanced SkipList Structure Optimization through the Use of PrevNode Mechanism
    Yeongyu Choi, Guangxun Zhao, Hojin Shin, Seehwan Yoo, Jongmoo Choi
  8. [KSC 2023] In-depth Analysis of Write Stall in LSM-Tree based Key-Value Store
    Guangxun Zhao, Hojin Shin, Yongjie Zhu, Jongmoo Choi
  9. [KSC 2022] Read performance analysis according to Compaction Trigger
    Sangwoo Kang, Guangxun Zhao, Min-guk Choi, Seehwan Yoo, Jongmoo Choi
  10. [KCC 2022] Retention Error Modeling using FEMU SSD Emulator
    Jeoungwon Lee, Guangxun Zhao, Jongmoo Choi
  11. [KSC 2021] Performance Analysis of BlobDB with Key-Value Separation Technique
    Jeoungwon Lee, Guangxun Zhao, Jongmoo Choi

Domestic Journals

  1. [정보과학회지 2026] LSM-tree 최적화를 위한 레벨별 SSTable 동적 크기 결정 기법
    Guangxun Zhao, JongMoo Choi
  2. [정보과학회지 2025] 학습된 인덱스가 키-밸류 스토어에 미치는 영향 분석
    Guangxun Zhao, JongMoo Choi

Research Studies

  • LSM-TreeJan–Feb 2025
  • Index StructuresJul–Aug 2024
  • FTLJul–Aug 2023
  • Linux KernelJan–Feb 2023
  • LevelDBJul–Aug 2022
  • RocksDBJul–Aug 2021

Honors & Awards

Best Research Award

Dankook University · Feb 2025

Top 3 research achievements among PhD and master's students in science.

Best Capstone Design Award

Dankook University · Dec 2021

1st place in the department capstone design competition.

Patent

System and Method for Replacing Bloom Filters with Learned Indexes in Key-Value Stores for Key Existence Detection

Korea · May 2025

Topics: LSM-tree, learned index, Bloom filter.

Certifications

  • Craftsman Information Equipment OperationKorea
  • BaristaKorea

Contact

For research discussions and collaboration, please contact me by email.