Jisang (Jace) Park
Research Scientist
My passion for social impact led me to found three startups, and it now motivates me to tackle South Korea’s labor shortage caused by the world’s lowest fertility rate. My solution is to expand robotics beyond industrial applications into everyday environments where natural and intuitive interaction with humans is required. To realize this, I aim to develop large multimodal models (LMMs) integrating visual, auditory, and linguistic cues for seamless human-robot interaction (HRI). Ultimately, I seek to realize my vision through entrepreneurship to create a worldwide impact beyond Korea.
- Location
- School of Computing, KAIST, Daejeon, South Korea
- jisang.park@kaist.ac.kr
Education
–
Seoul National University
Bachelor of Science in Computer Science & Engineering
Bachelor of Business Administration
- Summa Cum Laude (GPA: 3.9/4.0)
- Three-year full-ride scholarship from Bang-Il Young Foundation (highest GPA in the department)
Research Experience
– present
Korea Advanced Institute of Science and Technology (KAIST)
Research Scientist
Robust Intelligence & Robotics Lab (Advisor: Prof. Daehyung Park)
Highlights
- Achieved open-set semantic state estimation with 122% higher accuracy compared to closed-set baselines by developing a prompting method that grounds LMMs in affordances and temporal context
- Delivered 5 successful language-guided navigation demonstrations by implementing a real-time voice command framework through the integration of NVIDIA Riva ASR engine with ROS and RViz plugin
- Contributed to the publication of a survey featuring over 100 papers on the integration of LLMs into robotics by proposing the taxonomy and writing the communication section
– present
Independent Research
Independent Researcher
Remote Collaboration (Collaborator: Dr. Jongha Lee, Research Scientist @ Axoft Inc., Massachusetts)
Highlights
- Improved second-language (L2) pronunciation accuracy by 20% with only 10 minutes of training by developing an algorithm that approximates L2 phonemes with composite sounds of native phonemes
- Deployed a publicly accessible web service for pronunciation training through the development of a system that integrates the algorithm with FastAPI backend and React-based grapheme visualization
–
Seoul National University
Independent Researcher
Human Factors Psychology Lab (Advisor: Prof. Sowon Hahn)
Highlights
- Improved emotion recognition accuracy by up to 10% through augmentation of the MELD dataset with facial expression data and implementation of a multimodal model with a fusion of bcLSTM and VGG16
- Shaped collection guidelines for OPLEA dataset of 2.3K dialogue by analyzing the correlation between dialogue patterns and participant ratings and identifying reciprocity as a key engagement factor
Publications
Affordance and Scene Graph Prompting for Open-Set Semantic State Estimation with Large Multimodal Models
To be submitted to IEEE Robotics and Automation Letters (RA-L)
Graph-based prompting method to enable task-relevant and consistent estimation of semantic states in open-set settings
Inter-linguistic Phonetic Composition (IPC): A Theoretical and Computational Approach to Enhance Second Language Pronunciation
Arxiv Prepreint (under review in NAACL 2025)
Computational method to approximate L2 phonemes with composite sounds of native phonemes for enhanced pronunication
A survey on integration of large language models with intelligent robots
Intelligent Service Robotics
Review of LLM applications within core robotics elements—communication, perception, planning, and control
Feels like I've known you forever": empathy and self-awareness in human open-domain dialogs
Proceedings of the 44th Annual Conference of the Cognitive Science Society
Interdisciplinary study and collection of 2,300 human open-domain dialogs with self-labeled psychological variables
Entrepreneurial Experience
–
GreatZipsa Inc.
Co-Founder, CEO
Fintech Startup for Short-Term Rentals
Highlights
- Maximized return on investment to 23.3% by developing a rent optimization model encompassing both regression analysis of willingness-to-pay and sensitivity analysis of market factors
- Achieved a 99.7% annual occupancy rate by optimizing rental periods and timing through a time series analysis of 30K+ crawled online community posts for supply and demand estimation
- Selected for $45,000 government funding by Fintech Center Korea (top 10% among over 200 applicants) through an investment pitch highlighting technological innovation and social impact
–
Modoom
Co-Founder
Social Media Startup for Online Social Gathering
Highlights
- Acquired 1,000+ SNU student users organically and hosted 100+ gatherings within 2 weeks of launch by refining service based on 60+ user interviews and funnel analysis using Google Analytics and SQL
- Attracted 250+ students and achieved a 90% attendee satisfaction rate by hosting the SNU Online Christmas Party on a platform developed in 3 days by integrating Django with the Zoom API
- Hosted 2021 online freshmen orientations by establishing a partnership with the SNU Student Council
- Selected as one of 3 teams for SNU’s startup incubation program by demonstrating product-market fit
–
Enactus Seoul National University
Project Founder
Social Venture for Psychotherapy Services by Blind Counselors
Highlights
- Laid the foundation for a mental healthcare service that served 850+ customers and secured $105,000 in funding by developing affordable and accessible psychotherapy for depressed and lonely individuals
- Empowered 6 blind individuals with sustainable income and professional skills by recruiting, training, and employing them as counselors with a tailored curriculum co-developed with a psychotherapist
- Led Team Korea to the semi-finals in the Enactus World Cup among 36 competing countries as a presenter, highlighting the social impact of our projects to over 200 international attendees
Work Experience
–
Boston Consulting Group
Business Analyst
Highlights
- Validated commercial viability of new battery technology by conducting a correlation analysis between technical specifications and sales based on data collected for 300+ electric vehicle models
- Formulated an electric vehicle battery market entry strategy through quantitative analysis of government subsidies and manufacturer investment plans for supply-demand projections
–
Republic of Korea Army
Sergeant & Security Specialist
Highlights
- Awarded the Creativity Award at the ROKA Startup Competition for developing a dynamic password system using multi-factor authentication and PBKDF2 to enhance security and convenience (Top 5%)
- Reduced soldier workload by 60% and received a Letter of Commendation from the Division Commander for spearheading the surveillance system restructuring by leading stakeholders to consensus
Awards
Government Startup Funding Recipient from Fintech Center Korea
Selected for $45,000 government funding (top 10% among over 200 applicants)
Startup Incubation Program Selection from SNU Center for Entrepreneurship & Innovation
Selected as one of 3 teams for SNU’s startup incubation program
Startup Competition Creativity Award from Republic of Korea
Awarded for developing a dynamic password system (Top 5%)
Division Commander’s Letter of Commendation from Republic of Korea Army
Received for spearheading the surveillance system restructuring (Sole recipient)
Enactus World Cup Semi-Finalist from Enactus
Led Team Korea to the semi-finals among 36 competing countries as a presenter
Skills
- Language
- Keywords:
- Programming
- Keywords:
- Robot Platforms
- Keywords:
