Seongho Kim

Ph.D. in Electrical and Electronic Engineering

Yonsei University, Seoul, Republic of Korea

• Compiler Research Laboratory (CoreLab), advised by Prof. Hanjun Kim
• Full scholarship

B.S. in Electrical and Electronic Engineering

Yonsei University, Seoul, Republic of Korea

• Yonsei University Semiconductor Specialist Track

Research Assistant

Compiler Research Laboratory (CoreLab), Yonsei University

Advised by Prof. Hanjun Kim

• Developed Recifhe, a multi-level compiler that performs ciphertext- and polynomial-level optimization for fully homomorphic encryption (arXiv 2026)
• Designed a tensor abstraction that enables explicit ciphertext layout optimization for RNS-CKKS homomorphic encryption (CGO 2026 SRC)
• Co-developed a compiler-runtime cooperative verification framework for LLM-based code optimization (CGO 2026)
• Designed a multiplication algorithm for fully homomorphic encryption

Undergraduate Research Assistant

Compiler Research Laboratory (CoreLab), Yonsei University

Advised by Prof. Hanjun Kim

• Profiled kernel-level execution patterns of privacy-preserving machine learning workloads

Undergraduate Research Lead

Design and Architecture for Intelligent Systems Laboratory, Yonsei University

Advised by Prof. Joon-Sung Yang

• Analyzed inference bottlenecks across LLM architectures through systematic kernel profiling (OJCS 2025)

Undergraduate Research Intern

Scalable Computer Architecture Laboratory, Seoul National University

Advised by Prof. Jung Ho Ahn

• Analyzed the architecture of Llama 2 7B and identified optimization opportunities for inference acceleration

Undergraduate Research Intern

Circuit and System Laboratory, Yonsei University

Advised by Prof. Tae Wook Kim

• Designed a pulse generator for an ultra-wideband (UWB) transmitter

Information Systems Officer (Mandatory Military Service)

Headquarters, Republic of Korea Air Force

Emerging Engineering Researcher

Institute for the Promotion of Engineering and Science in Korea (IPESK)

Bronze Prize, 32nd Samsung Humantech Paper Award[Humantech]

Samsung Electronics

Compiler-Runtime Co-operative Chain of Verification for LLM-Based Code Optimization. Hyunho Kwon, Ju Min Lee, Hoyun Youm, and Seongho Kim.

Emerging Engineer

Institute for the Promotion of Engineering and Science in Korea (IPESK)

Edge LLM Inference Acceleration[GitHub]

SAIT Samsung Computer Engineering Challenge 2024 · Compiler Research Laboratory (CoreLab), Yonsei University

• Dynamic batching optimization framework for Microsoft Phi-3 on NVIDIA Jetson Orin (32 GB)
• Ranked 7th of 47 teams
• Architectural insights from the LLM bottleneck analysis (IEEE OJCS 2025)

Small Language Model Inference Acceleration[GitHub]

Backdrop Build V3 and V4, Backdrop Labs · Invited Project Builder

• Inference acceleration of Gemma 2B (Google DeepMind) with TensorRT-LLM

Session Volunteer

58th IEEE/ACM International Symposium on Microarchitecture (MICRO 2025), Seoul, Republic of Korea