Yunsung Lee¶
AI/ML Research Engineer
Leading research initiatives in Embodied AI, autonomous agents, and robotics foundation models. Currently Head of Research at MaumAI's WoRV team.
What You'll Find Here¶
👨💻 About Me
Learn about my background, research interests, and what drives my passion for AI technology.
📋 Professional CV
Comprehensive overview of my experience, publications, and achievements in AI research.
📝 Technical Blog
Insights, tutorials, and thoughts on AI research, machine learning, and technology trends.
🎯 Core Expertise¶
🤖 Vision-Language(-Action) AI
Expanding multimodal AI: vision → language → action
🎨 Diffusion Models
Generative AI and advanced diffusion-based approaches
🚀 Embodied Agents
AI systems that perceive and act in real-world environments
🏗️ Agentic AI
Autonomous AI agents with reasoning and decision-making capabilities
📚 Research Highlights¶
Recognition
Selected as an ICML 2026 Gold Reviewer — placed among top reviewers based on area-chair ratings, with complimentary conference registration as recognition.
Research Impact
My publications have accumulated over 1,851 citations on Google Scholar, demonstrating significant influence in machine learning and computer vision research.
Recent Publications¶
- Choosing the Right Prediction Target Maintains Realistic Manifold in Diffusion Models with Training-Free Guidance (Under Review)
- CostNav: A Navigation Benchmark for Real-World Economic-Cost Evaluation of Physical AI Agents (ScaleBot @ CVPR'26)
- WorldCam: Interactive Autoregressive 3D Gaming Worlds with Camera Pose as a Unifying Geometric Representation (VideoWorldModel @ CVPR'26)
- vla-eval: A Unified Evaluation Harness for Vision-Language-Action Models (From Data to Decisions @ ICRA'26)
- D2E: Scaling Vision-Action Pretraining on Desktop Data for Transfer to Embodied AI (ICLR'26)
- Multi-Architecture Multi-Expert Diffusion Models (AAAI'24)
💼 Technical Skills¶
📬 Get In Touch¶
I'm always interested in connecting with fellow researchers and discussing new ideas in AI and machine learning.