CV
AI research engineer building production-grade multimodal agents and generative AI systems. Takes initiative in identifying technical challenges and developing solutions with researchers, engineers, and product teams, with hands-on contributions across model development, evaluation, inference optimization, and deployment.
Contact Information
| Name | Jihoon Kim |
Experience
-
2023 - 2026 Seoul, Republic of Korea
AI Research Engineer
Cinamon
- Took initiative in shaping research priorities and contributing to technical reviews and delivery as the AI team scaled, working with researchers, engineers, motion specialists, and product teams to turn research into production-ready outcomes.
- Proposed and productized an agentic LLM/VLM system that coordinated multimodal reasoning and tool use to convert creative intent into precise shot specs and editable 3D previz, ensuring reliable execution, evaluation, and observability in production.
- Built a reusable generative-motion lifecycle spanning 3D data preparation, distributed model training, quantitative and expert evaluation, inference optimization, and production deployment.
- Built and maintained a reusable 3D data-processing package that transformed heterogeneous assets into validated, training-ready representations, establishing the data foundation for the company’s AI model development and production workflows.
- Took an active role in the research and publication process for a team paper as corresponding author, collaborating with coauthors on research direction, experimental design, technical review, and manuscript development.
-
2022 - 2022 Seongnam, Republic of Korea
Research Scientist Intern
KakaoBrain
- Conducted research and experiments across generative modeling approaches, including GANs, flow-based models, diffusion models, and text-to-image generation, covering model design, training, and evaluation.
- Conceived and conducted a research project on variable-length, language-conditioned motion generation, covering the full research lifecycle as first author, from problem formulation and model development to experimentation, evaluation, and manuscript submission.
-
2017 - 2020 Seoul, Republic of Korea
Machine Learning Scientist
AILYS (formerly Solidware)
- Delivered 10+ on-site machine-learning engagements across Korea, Japan, Vietnam, and Malaysia, working directly with client stakeholders to translate business objectives into well-scoped predictive-modeling problems and technical requirements.
- Developed and validated predictive models against client-defined criteria, iterating with stakeholders to align technical outcomes with operational needs.
- Researched, developed, and maintained an AutoML platform, translating machine-learning research and recurring customer requirements into reusable product capabilities.
- Enabled adoption by training business users at client organizations to use the solution independently in their day-to-day workflows.
-
2015 - 2017 Seoul, Republic of Korea
Corporate Property and Casualty Underwriter
Hanwha General Insurance
- Underwrote personal and corporate property-insurance products for Hanwha Group affiliates, evaluating property exposures and policy risk.
Education
-
2021 - 2023 Seoul, Republic of Korea
M.S.
Korea University
Artificial Intelligence
- GPA: 4.38 / 4.50; advised by Sungjoon Choi.
- Research: generative models, vision models, deep learning and its applications, human motion generation, and kinematics.
-
2010 - 2016 Seoul, Republic of Korea
B.S.
Hanyang University
Electronic Engineering
- GPA: 4.08 / 4.50 cumulative; 4.13 / 4.50 major; Summa Cum Laude.
-
2014 - 2015 Philadelphia, U.S.
International Exchange
Drexel University
Electrical and Computer Engineering
- GPA: 3.87 / 4.00; Dean’s List in all enrolled quarters.
Publications
-
Oct 2022 Learning Joint Representation of Human Motion and Language
arXiv
Jihoon Kim, Youngjae Yu, Seungyoun Shin, Taehyun Byun, Sungjoon Choi · Paper
Patents
- Method for applying user intension to unsupervised learning and apparatus therefor · Jun 2021
Korean Patent No. 102273868
- Method and apparatus for generating supervised learning model based on unsupervised learning, and method and apparatus for analyzing unsupervised learning model using the supervised learning model · Jun 2021
Korean Patent No. 102273867
- Method and apparatus for automatically generating variables for data modeling · May 2019
Korean Patent No. 101976689
Skills
Honors and Awards
-
2018 Bronze Award
Kyobo Insuretech Hackathon, Kyobo Insurance
-
2017 Bronze Award
Culture and Tourism Big Data Contest, Korea Culture & Tourism Institute
-
2015 First Prize
Engineering School Capstone Design Fair and Graduation Project Competition, Hanyang University
-
2014 Qualcomm Scholarship Program
Qualcomm Korea
-
2014 Best Tutor Award
Engineering Mathematics Tutoring Program, Hanyang University
-
2014–2015 National Science & Technology Scholarship
KOSAF
-
2010–2015 Academic Excellence Award
Hanyang University