Jeonggyu Kang

Master of Science in Computer Science

Korea Advanced Institute of Science and Technology (KAIST)

Collaborative Social Technologies Lab · Advisor: Prof. Joseph Seering

Bachelor of Science in Electrical Engineering

Korea Advanced Institute of Science and Technology (KAIST)

GPA 3.88/4.3 (95.80/100) · Magna Cum Laude

research

Graduate Researcher

KAIST Collaborative Social Technologies Lab · Advisor: Prof. Joseph Seering

Human-centered NLP and multi-agent systems

• Co-led TIDES, a longitudinal bilingual corpus of 12 teams, 104 meetings, and 76K+ utterances, annotated with speaker roles, developmental stages, and interaction-process categories (accepted to COLM 2026)
• Studying whether internal model controls can manufacture useful cognitive diversity within a single base model, building a systematic falsification record with token-matched baselines and effect-size analysis
• Investigating whether diversity survives extended multi-agent conversation without converging, using multi-turn debate among stance-seeded agents

industry

Intern, S/W Development Team

Samsung Electronics, Device Solution Division

Development of advanced LLM service enhancement project

• Developed PIR-based Query Decomposition Unit: 28% search quality improvement, 43% zero-hit query reduction
• Implemented Query-Document Relevance Check with DeepSeek-R1-Distill-Qwen-32B: 98.1% score stability
• Built RAG evaluation system applied to 3,000+ real-world evaluations
• Researched Korean-optimized LLM prompt engineering techniques

research

Independent Research

KAIST Integrated Vision Language Lab

Research on LLM and RAG system optimization

• Comparative research on standard vs reasoning-enhanced LLMs
• Investigated optimal retrieval methodologies for RAG systems
• Fine-tuning experiments for Korean-specific LLM performance

research

Independent Research

KAIST Neuro-Instrumentation & Computational Analysis Lab

Research on deep neural networks for computer vision

• In-depth study based on Stanford CS231n curriculum
• CNN architecture design and optimization
• Image segmentation, object detection, and image generation implementation
• Deep learning training pipelines with PyTorch