3D/4D Vision & Spatial Intelligence
Learning spatial and spatiotemporal representations for 3D reconstruction, dynamic-scene understanding, novel-view synthesis, and neural rendering.
HELLO, I'M
Jooho Son
I am an undergraduate researcher at KAIST working on 3D computer vision, neural rendering, reinforcement learning, and multi-agent systems.
I am interested in systems that understand, reconstruct, and act in the physical world. My work connects research ideas to reproducible experiments—clear assumptions, controlled evaluation, failure analysis, and working code.
I am currently involved in research at KAIST CVLab. Previously, I explored AI for Science at DSAIL and led applied AI, education, and consulting projects.
Research Interests
Vision-Language-Action (VLA)
Connecting visual perception, language reasoning, and embodied action to build agents that understand and interact with the physical world.
Human–AI Interaction & Multi-Agent Systems
Designing human-centered AI experiences and coordinated agent systems that support communication, collaboration, and real-world decision-making.
CVLab Publications & Ongoing Research
2026
4D Novel View Synthesis
KAIST CVLab collaborative research · Contributor (non-first author)
Contributed literature review, experimental-condition design, quantitative evaluation tables, baseline comparisons, and paper figures for dynamic-scene reconstruction research.
Manuscript under reviewOngoing
City-Scale Mesh Reconstruction
KAIST CVLab collaborative research · Contributor
Contributing prior-art and baseline analysis, problem formulation, pipeline design, and weekly research materials for large-scale urban 3D reconstruction.
Work in progressAdditional project pages, paper links, and code will be added as they become publicly available.
Selected Projects
Tech for Impact · 2025
Ieum ↗
Developed an AI assistant-powered communication service designed to help adolescents with developmental disabilities express themselves and engage in supportive conversations.
- Assistive AI
- Conversational UX
- Social Impact
Field Study · 2026
Starlink × Sky View
Measured the relationship between sky obstruction and satellite-internet performance by synchronizing terminal, GPS, network, and sky-mask data.
- Networking
- Measurement
- Data Analysis
Reinforcement Learning · 2026
Top-Gun RL
Developed a combat-simulation training workflow using PPO, SAC, curriculum learning, self-play, checkpoint recovery, and sequence replay.
- Self-play
- PPO / SAC
- Simulation
Multi-Agent Systems · 2026
Plan Bee
Built a proxy-agent negotiation system that replans work when schedule changes create bottlenecks in software teams.
- LangGraph
- FastAPI
- React
Activity
Selected moments from consulting, education, and applied AI work beyond the lab.
Strategy & Consulting · 2024–2025
KAIST Management Study Group (MSK)
Led and contributed to strategy projects for Frip, Nota AI, and Coupang Play, turning market research into concrete service and go-to-market proposals.
Global Volunteering · 2025
AI Education in Timor-Leste
Designed and delivered an English-language AI development curriculum with KAIST volunteers and students at the National University of Timor-Leste.
AI Competition · 2026
Hanwha Life Future Talent Contest
Won 3rd Prize in the AI sector by proposing a structured data pipeline and model-selection framework for an insurance AI solution, then presented the work at the final stage.
Vitæ
Download ResumeChungnam Science High School
29th Cohort · Graduated early before entering KAIST.
KAIST Management Study Group
Vice President · Led consulting projects for Nota AI and Frip, translating market questions into executable strategies.
Timor-Leste Global Volunteering Group
AI Education Lead · Designed and delivered an English-language AI development curriculum.
Previous research labKAIST DSAIL
Undergraduate Researcher · Reviewed AI for Science research and studied graph learning and scientific reasoning pipelines.
Hanwha Life Future Talent Contest
3rd Prize, AI Sector · Recognized for an AI project in Hanwha Life's future-talent competition before joining CVLab.
Current research labKAIST Computer Vision Lab (CVLab)
Undergraduate Researcher · Joined CVLab after the Hanwha Life contest, following the earlier research experience at DSAIL. Currently conducting research on 4D novel-view synthesis, city-scale mesh reconstruction, action latent representations, and vision-language-action models.
Skills
Certifications & Language
Craftsman Information Processing
Korean National Technical Qualification
APEX AI Certification
TOEIC 835
English proficiency
Technical Skills
Research: 3D Vision, Neural Rendering, RL, Multi-Agent Systems
Languages: Python, C, Java, Scala, TypeScript
Systems: PyTorch, Node.js, SQL, MongoDB, Git
Collaboration & Coffee Chats
I am always open to discussing research, projects, and new ideas.