Research Interests
I am an undergraduate researcher focusing on automated design space exploration across circuit and architectural levels for AI acceleration. My work examines algorithmic approaches for generating and validating hardware structures beyond manually designed templates, spanning behavioral state-transition representations, CMOS implementations, and accelerator organizations.
I am particularly interested in abstraction-aware representation design for hardware generation and optimization, as well as architecture-level design space exploration methods that derive accelerator organizations from workload and mapping characteristics.
Education
Ulsan National Institute of Science and Technology (UNIST)
2021 – 2027 (expected)Bachelor of Engineering in Electrical and Electronic Engineering
GPA: 4.22 / 4.30 (through Fall 2025)
Honors & Awards
Unhae Scholarship
UNIST · Merit-based scholarship (KRW 8,000,000)
Dean's Award for Academic Excellence
UNIST · EE Department · Department Top 1
Dean's Award for Academic Excellence
UNIST · EE Department · Department Top 1
New Student Division Award for Academic Excellence
UNIST · Freshman Top 3
Publications & Posters
Jeong, H.*, Kang, K.*, Jung, W., Lee, J.
"B-Flex: Exploration of Broader Flip-Flop Design Space Based on FSM Exhaustive Search"
Accepted to DAC 2026 (Design Automation Conference)
*Co-first authors
Selected Projects
B-Flex: Exploration of Broader Flip-Flop Design Space Based on FSM Exhaustive Search
Formalized behavioral equivalence between asynchronous FSM representations and flip-flop functionality and developed a pruning-based generative search enabling scalable exhaustive exploration of the FF design space, discovering over 568 million valid mechanisms and synthesizing novel CMOS flip-flop topologies with improved performance characteristics.
Key Highlights
- •Discovered 568M+ valid FF FSM mechanisms through exhaustive 3-bit-state exploration
- •Identified novel FF designs achieving 2.53× speedup and improved metastability over transmission-gate FFs
- •Resolved AFSM validation incompleteness via complete graph-based behavioral equivalence checking
MICA: Multimodal Interactive Conversational Agent
Undergraduate Research Program
Contributed to the development of a Korean multimodal conversational agent integrating KoBERT text encoders with MediaPipe-based facial cues; participated in video dataset curation and model training for context-aware response generation.
Technical Skills
Languages
C++, Python, Verilog
Dev Tools
Git, Docker
Utility / Collaboration
Figma, draw.io