Dohyun Kim

About me

I'm an engineer and researcher working on 3D perception and AI research. I studied mechanical and electrical engineering at KAIST, and earned my M.S. with research on stereo vision-based 3D measurement. These days at Recon Labs I rebuild real scenes as neural representations and build AI agents that help people create.

What connects all of it: turning images into something you can trust about the world — a coordinate, a shape, a decision.

[ Curriculum Vitae ]

Dohyun Kim

01

Projects

3D Gaussian Splatting R&D,
Recon Labs, 2024 — 2025

Neural 3D Reconstruction
Stereo camera measurement setup between ship blocks at a shipyard — three tripod-mounted cameras and a checkerboard target

Ship-block stereo measurement,
HUMAN Lab KAIST, 2022 — 2023

Stereo Vision, Calibration
Ball-collecting robot NUNU — the built robot with its roller pick-up module

Ball-collecting robot,
KAIST ME Undergraduate Project, 2019

RGB-D Vision, Path Planning, Mechanical Design
Node-graph creative canvas with an AI agent side panel searching canvas nodes by content — agent harness built at Recon Labs

Agent harness for a creative canvas,
Recon Labs, 2025 — 2026

AI Agent, Orchestration
Pencil sketch of Notre-Dame de Paris with a street scene — tour bus, cyclists, and cobblestone road in the foreground

Pencil on Paper, 2017

Urban Sketch
claw-hwp building a government-style bar chart inside an HWP document, animated capture

claw-hwp — HWP document automation skill,
Personal Open-source Project, 2026

Claude Agent Skill, Document Automation

02

Experience

Mar. 2024 — Present

AI Researcher

Recon Labs, Seoul

  • Researched 3D Gaussian Splatting for a production 3D capture service — reconstruction quality for real-world captures, artifact and floater suppression, background modeling, and rendering optimization.
  • Developed virtual try-on features for a fashion service by combining and evaluating multiple generative models.
  • Built an AI agent and its orchestration harness for a graph-based generative content platform.

Mar. 2022 — Mar. 2024

Graduate Researcher

HUMAN Laboratory, KAIST, Daejeon

  • Developed stereo vision-based measurement workflows for industrial-scale structures — camera calibration and image preprocessing for high-speed, high-precision 3D measurement.
  • Built automatic correspondence-point detection using feature keypoints, ROI separation, and co-visible region analysis.
  • Contributed signal-processing algorithms and CNN-based extensions for a compact optical 3D sensing platform.
  • Worked on RGB-D camera-array positioning and motion recognition for an XR ship-steering training system.

03

Education

Mar. 2022 — Mar. 2024

KAIST

M.S., Mechanical Engineering

  • Graduate research on stereo matching, camera geometry, and 3D perception at HUMAN Laboratory.
  • Coursework includes: Visual Intelligence, Autonomous Mobile System, Machine Learning for 3D Data, Probability and Statistics.

Mar. 2015 — Jun. 2021

KAIST

B.S., Mechanical Engineering · Double Major in Electrical Engineering

  • Foundations across mechanics, circuits, signals, and control — the habit of thinking in systems.

04

Contact Me

Email

kbh446688@gmail.com

Social Network