Hello, I'm a/an

AI Agents + Safety + Health

I'm Yubin, a research scientist at Google DeepMind and a PhD candidate at the MIT Media Lab .
I build reliable, scalable AI agents that collaborate with humans and robots to accelerate scientific discovery and deliver personalized, long-term health support in the real world.

Media & Talks

TechXplore · Sep 2026

Why Adding AI Agents Sometimes Reduces Performance

AI Talks with NTU & NUS · Feb 2026

Towards a Science of Scaling Agent Systems

MBZUAI AI Nexus · Dec 2025

Healthcare Agents: Language Models in Health Prediction and Decision-Making

NeurIPS 2024 Oral · Dec 2024

MDAgents: Adaptive Collaboration of LLMs for Medical Decision-Making

Life Updates

2026

Aug Our paper on "MatrAIx: Simulating the World with 8.3 Billion Persona Agents" is released as a preprint paper.
Aug Earned the 2026 MIT Professional Development Certificate in Research Mentoring.
Jul Our paper on "Capable language models can outgrow the benefits of collaboration" is published in Nature Machine Intelligence.
May Our paper on "Towards a General Intelligence and Interface for Wearable Health Data" is released as a preprint paper
May Our paper on "TeamBench: Evaluating Agent Coordination under Enforced Role Separation" is released as a preprint paper
May I've started a student researcher position at Google DeepMind
Apr Our paper on "Social Human Robot Embodied Conversation (SHREC) Dataset: Benchmarking Foundational Models' Social Reasoning" is accepted at RSS 2026
Apr Our paper on "CoDaS: AI Co-Data-Scientist for Biomarker Discovery via Wearable Sensors" is released as a preprint paper
Mar Gave an invited talk on "Towards a science of scaling agent systems" at 🎙️.
Feb Gave an invited talk on "Towards a science of scaling agent systems: When and why agent systems work?" at the AI Talks with NTU & NUS. I enjoyed the discussions that followed, especially on how we can coordinate agents with different capabilities and the trade-offs between capability and cost 🎙️.
Jan Gave an invited talk on "Agent Scaling Law" at Adobe Research . I enjoyed the insightful discussions that followed, especially on how we can approach evaluation for agent systems and architectural design 🎙️.
Jan Our paper on "Towards a Science of Scaling Agent Systems" is released as a preprint paper

2025

Dec Gave an invited talk on "Healthcare Agents: Language Model Agents in Health Prediction and Decision-Making" at MBZUAI 🎙️.
Sep Our paper on "Tiered Agentic Oversight: A Hierarchical Multi-Agent System for AI Safety in Healthcare" is selected as a spotlight poster at AI Cures 2025.
Aug Our paper on "The Anatomy of a Personal Health Agent" is released as a preprint paper
Aug Our paper on "BehaviorSFT: Behavioral Token Conditioning for Clinical Agents Across the Proactivity Spectrum" is accepted as a findings paper at EMNLP 2025.
Jun I've started a research intern program at Google Research
May Our paper on "VocalAgent: Large Language Models for Vocal Health Diagnostics with Safety-Aware Evaluation" is accepted as an oral paper at Interspeech 2025.
Mar Our paper on "Social robots as conversational catalysts: Enhancing long-term human-human interaction at home" is published as a cover paper at Science Robotics.

2024

Sep Our paper on "MDAgents: An Adaptive Collaboration of LLMs for Medical Decision-Making" is accepted as an oral paper (Top 0.39%) at NeurIPS 2024.
Jul Our paper on "EmpathicStories++: A Multi-modal Dataset for Empathy towards Personal Experiences" is accepted as a Findings at ACL 2024.
May I've started a research internship at Honda Research Institute on LLMs in Displaying Group Dynamics
Apr Our paper on "Health-LLM: Large Language Models for Health Prediction via Wearable Sensor Data" is accepted at CHIL 2024.
Mar Our paper on "Diffusion-EDFs: Bi-equivariant Denoising Generative Modeling on SE(3) for Visual Robotic Manipulation" is accepted at CVPR 2024.

2023

Sep I started the MAS graduate program at MIT Media Lab
Jul Our paper on "HIINT: Historical, Intra-and Inter-personal Dynamics Modeling with Cross-person Memory Transformer" is accepted at ICMI 2023
Apr Our paper on "Multipar-T: Multiparty-Transformer for Capturing Contingent Behaviors in Group Conversations" is accepted at IJCAI 2023
Jan Our paper on "Joint Engagement Classification using Video Augmentation Techniques for Multi-person Human-robot Interaction" is accepted at AAMAS 2023

2022

Jul I started my visiting scholar program at MIT

Portfolio | Yubin Kim

Projects

My experience has been focused on implementing personalized technology with foundation models for human-agent interactions in an intuitive and reciprocal manner. Specifically, I am interested in developing personal agents that can maintain long-term interactions with humans in the wild and perform continual learning.

CoDaS: AI Co-Data-Scientist for Biomarker Discovery via Wearable Sensors
Under Review

TeamBench: Evaluating Agent Coordination under Enforced Role Separation
Under Review

Towards a General Intelligence and Interface for Wearable Health Data
Under Review

Tiered Agentic Oversight: A Hierarchical Multi-Agent System for AI Safety in Healthcare
AI Cures 2025 Spotlight

The Anatomy of a Personal Health Agent
Under Review

Towards a Science of Scaling Agent Systems
Nature Machine Intelligence

InvThink: Mitigating Safety Risks Through Inverse Reasoning
Under Review

Health-LLM: Large Language Models for Health Prediction via Wearable Sensor Data
CHIL 2024

MDAgents: Adaptive Collaboration Strategy for LLMs in Medical Decision Making
NeurIPS 2024 Oral (Top 0.39%)

Conversational AI Agents for Elder Care

Digital Companion Robot Design for Older Adults

Development of Auto Dyadic App

MineRL: Training AI agents in Open World Environments

ICRA RoboMaster AI Challenge

Development of Indoor Serving Robot

Explainable Human-Robot Interaction Learning with Video Augmentation Techniques
AAMAS 2023