Hi! I’m Chaemin Lim, a Ph.D. candidate at Yonsei University / HPCP Lab, advised by Prof. Youngsok Kim.

My research broadly spans systems software, with a primary focus on database systems. I design hardware-aware execution and data-movement mechanisms for heterogeneous platforms, ranging from GPU-, PIM-, and DPU-accelerated servers to resource-constrained and trusted edge devices.

My goal is to translate emerging hardware capabilities into practical, efficient, and secure end-to-end systems.

Research Interests

• Systems software, with an emphasis on database systems
• GPU-accelerated analytical and out-of-memory query execution
• Near-data processing with PIM and DPU hardware
• Join processing, skew handling, collective communication, and data movement
• Secure and resource-constrained edge computing, trusted execution environments, and on-device inference

Publications

1. GJP: Fast and Efficient Join-Project Query Processing on GPUs Using Density-Aware Matrix Multiplication
   Sungchul Lee, Suhyun Lee, Hanna Cha, Chaemin Lim, and Youngsok Kim.
   2027 ACM International Conference on Management of Data (SIGMOD), June 2027 (to appear).
   Keywords: GPU; Database Systems; Join-Project Processing; Matrix Multiplication
2. FaScalSQL: A Fast and Scalable GPU-Accelerated SQL Query Engine for Out-of-Memory Tables
   [GitHub]
   Chaemin Lim, Suhyun Lee, Jinwoo Choi, Kwanghyun Park, Jinho Lee, Joonsung Kim, and Youngsok Kim.
   42nd IEEE International Conference on Data Engineering (ICDE), May 2026.
   Keywords: GPU; Database Systems; Out-of-Memory Query Processing
3. DMO-DB: Mitigating the Data Movement Bottlenecks of GPU-Accelerated Relational OLAP
   Chaemin Lim, Suhyun Lee, Jinwoo Choi, Joonsung Kim, Jinho Lee, and Youngsok Kim.
   34th International Conference on Parallel Architectures and Compilation Techniques (PACT), Nov. 2025.
   Keywords: GPU; Database Systems; Data Movement; Relational OLAP
4. SPID-Join: A Skew-resistant Processing-in-DIMM Join Algorithm Exploiting the Bank- and Rank-level Parallelisms of DIMMs [GitHub]
   Suhyun Lee, Chaemin Lim, Jinwoo Choi, Heelim Choi, Chan Lee, Yongjun Park, Kwanghyun Park, Hanjun Kim, and Youngsok Kim.
   2025 ACM International Conference on Management of Data (SIGMOD), June 2025.
   Keywords: PIM; Database Systems; Join Processing; Data Skew
5. PID-Comm: A Fast and Flexible Collective Communication Framework for Commodity Processing-in-DIMMs [GitHub]
   Si Ung Noh¹, Junguk Hong¹, Chaemin Lim, Seongyeon Park, Jeehyun Kim, Hanjun Kim, Youngsok Kim, and Jinho Lee.
   51st IEEE/ACM International Symposium on Computer Architecture (ISCA), June–July 2024. ¹First Co-Author
   Keywords: PIM; Collective Communication; Distributed Systems
6. Design and Analysis of a Processing-in-DIMM Join Algorithm: A Case Study with UPMEM DIMMs [GitHub]
   Chaemin Lim, Suhyun Lee, Jinwoo Choi, Jounghoo Lee, Seongyeon Park, Hanjun Kim, Jinho Lee, and Youngsok Kim.
   2023 ACM SIGMOD International Conference on Management of Data (SIGMOD), June 2023.
   Keywords: PIM; Database Systems; Join Processing
7. GuardiaNN: Fast and Secure On-Device Inference in TrustZone Using Embedded SRAM and Cryptographic Hardware
   Jinwoo Choi¹, Jaeyeon Kim¹, Chaemin Lim¹, Suhyun Lee, Jinho Lee, Dokyung Song, and Youngsok Kim.
   23rd ACM/IFIP International Middleware Conference (Middleware), Nov. 2022. ¹First Co-Author
   Keywords: Trusted Execution Environments; On-Device Inference; Security

Education

• [Sep. 2023 - Present] Ph.D. Candidate in Computer Science, Yonsei University; advised by Prof. Youngsok Kim, High Performance Computing Platform (HPCP) Lab
• [Mar. 2021 - Aug. 2023] M.S. in Artificial Intelligence, Yonsei University; advised by Prof. Youngsok Kim, High Performance Computing Platform (HPCP) Lab
• [Mar. 2017 - Feb. 2021] B.A. in Journalism; B.Eng. in Computer Engineering (Double Major), Hankuk University of Foreign Studies

Patents

• Skew Resistance Processing in a DIMM Device and Processing Method. US Patent Application 18/933,770, 2026. Youngsok Kim, Suhyun Lee, Chaemin Lim, Jinwoo Choi, and Hanjun Kim.
• Electronic Device and Method with Processing in Memory Operation. US Patent Application 19/060,889, 2025. Bongjun Kim, Youngsok Kim, Sungchul Lee, Seil Lee, Suhyun Lee, Jounghoo Lee, Chaemin Lim, Hanna Cha, Jinwoo Choi, Yeonan Ha, and Hanwoong Jung.
• PIM-Based Computing Device for Accelerating Join Operation. US Patent 12,386,778, 2025. Youngsok Kim, Chaemin Lim, Hanjun Kim, and Jinho Lee.
• Artificial Intelligence Device Based on Trust Environment. US Patent Application 18/229,765, 2024. Youngsok Kim, Jinwoo Choi, Chaemin Lim, Suhyun Lee, Dokyung Song, and Jinho Lee.

Invited Talks & Conference Presentations

• FaScalSQL: A Fast and Scalable GPU-Accelerated SQL Query Engine for Out-of-Memory Tables. Conference Presenter, IEEE ICDE, May 2026.
• DMO-DB: Mitigating the Data Movement Bottlenecks of GPU-Accelerated Relational OLAP. Conference Presenter, IEEE/ACM PACT, Nov. 2025.
• A Fast Processing-in-DIMM Join Algorithm Exploiting UPMEM DIMMs. Invited Speaker, ABUMPIMP, EUROPAR, Aug. 2023.
• Design and Analysis of a Processing-in-DIMM Join Algorithm: A Case Study with UPMEM DIMMs. Conference Presenter, ACM SIGMOD, June 2023.

Awards

• [2024] Encouragement Prize, Samsung Humantech Paper Awards (삼성휴먼테크논문대상) — "PID-Comm: Fast and Flexible Collective Communication Framework for Commodity Processing-in-DIMM Devices" (Junguk Hong, Si Ung Noh, Seongyeon Park, and Chaemin Lim).

Employment

• [Aug 2025 - Jul 2026] System Architect Intern, Mangoboost [GitHub]
  • Worked primarily on system architecture and performance optimization for prefill–decode (PD) disaggregated LLM serving on AMD GPU clusters, focusing on RDMA-based KV-cache transfer, multi-node serving, and ROCm inference paths.
  • Improved the MoRIIO serving path across asynchronous KV-transfer completion and cleanup, send draining, READ handoff routing, and multi-node TP=16 prefill–decode execution.
  • Made selected upstream contributions to vLLM: AITER MLA/MTP and FP8 correctness, shared-memory broadcast reliability, and MoRIIO proxy health monitoring.

Funded Projects

• Development of New Concept PIM Semiconductor Leading Technology — IITP, 2022–Present
  • Developed database and communication systems on commodity UPMEM DIMMs, including PID-Join, SPID-Join, and PID-Comm.
  • Released reproducibility artifacts for join processing and collective communication.
• STACK: Smart, Attack-Resistant Internet of Things Networks — KIAT, 2021
  • Worked with trusted execution environment software for attack-resistant IoT systems.
  • Reported an STM32MP1 clock-rate computation bug in OP-TEE OS; the merged fix includes a Reported-by acknowledgment. OP-TEE #4904

Teaching Experience

• [Fall 2022] Teaching Assistant, Logic Circuit Design (CSI2111), Yonsei University
• [Fall 2021] Teaching Assistant, Open Source Projects (CSI2115), Yonsei University
• [Spring 2021] Teaching Assistant, Information Security (CSI4109), Yonsei University

Technical Skills

Programming: C, C++, Python, Scala, Rust, SQL

Misc

• I believe every useful computational device ever built can run DOOM.