HCI Research Scientist · TwelveLabs

Taewan Kim, Ph.D.

I research and design how people interact with AI to understand themselves and the world.

I'm an HCI researcher and designer working at the intersection of human-AI interaction and AI-driven service design. At TwelveLabs I design the human-AI interaction language for video understanding. Before that I was a Senior Designer at Samsung Electronics and a research intern at NAVER AI Lab. I earned my Ph.D. from KAIST, advised by Professor Hwajung Hong, where I studied how large language models can support reflection and mental wellbeing in clinical and everyday settings.

twkim24@gmail.com
Google Scholar CV LinkedIn

News

Jockey Research Preview publicly launched! I contributed the interaction language and surface design for human-video AI interaction.

Paper on LLM-generated patient-friendly clinical notes published in Scientific Reports.

Joined Twelve Labs as an HCI Research Scientist, focusing on human-AI interaction and multimodal video understanding.

Paper on exploring design spaces to facilitate household collaboration accepted to CHI 2025.

Invited talk at Korean Neuropsychiatric Association Spring Conference.

Experience

  • TwelveLabs

    HCI Research Scientist · San Francisco / Seoul

    Researching human-AI interaction for video AI, and designing interaction flows that make model outputs adaptive, interpretable, and part of real workflows.

  • Samsung Electronics

    Senior Designer · Seoul

    AI Creator in Samsung's generative AI design program. Explored synthetic persona-based user studies for early-stage UX design.

  • NAVER AI Lab

    Research Intern · Seongnam

    Led research on large language models for mental health support, with Dr. Young-Ho Kim.

Education

  • KAIST

    Ph.D. in Industrial Design · Advisor: Prof. Hwajung Hong

    Dissertation: Designing for Mental Wellbeing: Exploring the Role of AI for Reflective Practice

  • Seoul National University

    Ph.D. candidate in Cognitive Science

    Completed coursework, then moved to KAIST with my advisor.

  • UNIST

    M.S. in Creative Design Engineering

  • Handong University

    B.S. in Product Design

Off Work

Research

Research Vision & Framing

Designing AI as an Interactive Layer for Human Sensemaking

AI is becoming the layer through which people perceive, navigate, and act in the world. It is no longer just generating static information, but increasingly shaping human interpretation through interaction.

AI mediates how humans understand themselves.

Exploring how AI can support mental well-being through everyday interaction and reflection.

MindfulDiary journaling app screens
CHI '24 MindfulDiary ↗
DiaryMate journaling interface
CHI '24 DiaryMate ↗

AI mediates how humans understand the world.

Exploring how AI can help people navigate and interpret complex, multimodal information such as video.

Jockey by TwelveLabs
Research Preview Interaction Language for Jockey ↗

Jockey is TwelveLabs' agentic video AI. I also contributed its surface design.

Selected Work

  • 2026 · TwelveLabs

    Interaction Language for JockeyAgentic video AI, research preview

    Jockey is TwelveLabs' agentic video AI, which organizes, searches, and assembles across an entire video library. I contributed its interaction language and surface design.

  • CHI 2024

    MindfulDiaryLLM journaling for psychiatric patients

    A mobile journaling app designed with mental health professionals, which keeps free-form LLM conversation within clinicians' guidelines. In a four-week field study with 28 patients with major depressive disorder and five psychiatrists, patients kept richer daily records and psychiatrists understood their patients' daily contexts better.

  • CHI 2024

    DiaryMateWriting a diary together with an LLM

    A technology probe that suggests the next sentence of a journal entry. A ten-day field study with 24 people showed that AI suggestions helped them see their own experiences from several perspectives, and that some came to rely on the AI's wording for their emotions.

  • CHI 2022

    MindScopeAlgorithmic stress prediction for self-reflection

    A personal informatics system that predicts stress and explains its prediction. In a 25-day field study with 36 college students, the predictions and explanations prompted people to revisit their assumptions about their own stress and plan how to cope. Featured in KAIST Breakthroughs.

  • CHI 2020

    A Vulnerable BotA social bot that receives care

    A bot that asks for comfort instead of offering it. Over three weeks, 55 college students who responded to it reframed their own negative experiences, reported less stigma toward depression, and grew more open to seeking treatment.

Publications

Journal & Conference Papers

Extended Abstracts & Workshop Papers