Hyunseung Kim

I am an AI Researcher at KRAFTON AI, mentored by Kangwook Lee. I'm building embodied agents that can reason, adapt, and act autonomously in complex, real-time environments. Also, I am a Ph.D. candidate at KAIST, advised by Jaegul Choo.

Email / CV / Google Scholar / Linkedin / Github

News

06.2026. PUBG Ally is released to the public as a beta service!
06.2026. 3D HAMSTER is accepted to IROS 2026!
02.2025. - Present. AI Companion Team Lead at KRAFTON AI.
01.2025. SimBa is accepted to ICLR 2025 as a spotlight presentation!
09.2024. DoDont is accepted to NeurIPS 2024!
08.2024. - 02.2025. AI Research Internship at KRAFTON AI.
05.2024. Two papers, Hare&Tortoise and CoIn are accepted to ICML 2024!
09.2023. Two papers, DISCO-DANCE and PLASTIC are accepted to NeurIPS 2023!

Experience

KRAFTON AI Aug. 2024 - Present

AI Researcher · AI Companion Team Lead

PUBG Ally Project Project Lead

PUBG Ally is the world's first in-game AI agent that can reason, act autonomously, and play alongside players as a voice-enabled teammate in a live battle royale game. Led the project end to end to reach live service: agent architecture, data collection, training, evaluation, and on-device optimization.

post / video / many more

Talks

(July 2026) CDEC 2026, Shanghai — From NPCs to Teammates: Building the Next Generation of Game AI
(June 2026) Game AI Reading Group, Sony AI, online — PUBG Ally: Building a Co-Playable AI Teammate in PUBG
(Mar. 2026) GDC 2026, San Francisco (video) — Building a Co-Playable Character: PUBG Ally, an AI Teammate Powered by NVIDIA ACE (with Evgeny Makarov, NVIDIA)
(Oct. 2025) HCAI Research Workshop, SNU-CMU Human-Centered AI Research Center, Seoul — Building Co-Playable Characters in Games

Publications

3D HAMSTER: Bridging Planning and Control in Hierarchical Vision Language Action Models through 3D Trajectory Guidance
Dongyoon Hwang*, Byungkun Lee*, Dongjin Kim*, Hyojin Jang, Hoiyeong Jin, Jueun Mun, Minho Park, Hojoon Lee, Hyunseung Kim, Jaegul Choo
IROS'26.
project page / paper / code

PHUMA: Physically Reliable Humanoid Locomotion Dataset
Kyungmin Lee†, Sibeen Kim†, Youngdo Lee, Minho Park, Hyunseung Kim, Dongyoon Hwang, Donghu Kim, Hojoon Lee, Jaegul Choo
Preprint.
project page / paper

SimBa: Simplicity Bias for Scaling Up Parameters in Deep Reinforcement Learning
Hojoon Lee*, Dongyoon Hwang*, Donghu Kim, Hyunseung Kim, Jun Jet Tai, Kaushik Subramanian, Peter R. Wurman, Jaegul Choo, Peter Stone, Takuma Seno.
ICLR'25 (Spotlight).
project page / paper

Do’s and Don’ts: Learning Desirable Skills with Instruction Videos
Hyunseung Kim, Byungkun Lee, Hojoon Lee, Dongyoon Hwang, Donghu Kim, Jaegul Choo
NeurIPS'24.
project page / paper

Slow and Steady Wins the Race: Maintaining Plasticity with Hare and Tortoise Networks
Hojoon Lee, Hyeonseo Cho, Hyunseung Kim, Donghu Kim, Dugki Min, Jaegul Choo, Clare Lyle
ICML'24.
paper / code

A Simple Convolution INjector for Vision Transformer: Towards Effective Adaptation in Visuo-Motor Control
Dongyoon Hwang*, Byungkun Lee*, Hojoon Lee, Hyunseung Kim, Jaegul Choo
ICML'24.
paper

DISCO-DANCE: Learning to Discover Skills through Guidance
Hyunseung Kim*, Byungkun Lee*, Hojoon Lee, Dongyoon Hwang, Sejik Park, Kyushik Min, Jaegul Choo
NeurIPS'23.
project page / paper / code / poster

PLASTIC: Improving Input and Label Plasticity for Sample Efficient Reinforcement Learning
Hojoon Lee*, Hanseul Cho*, Hyunseung Kim*, Daehoon Gwak, Joonkee Kim, Jaegul Choo, SeYoung Yun, Chulhee Yun
NeurIPS'23.
paper / code / slide / poster

DraftRec: Personalized Draft Recommendation for Winning in MOBA Games
Hojoon Lee*, Dongyoon Hwang*, Hyunseung Kim, Byungkun Lee, Jaegul Choo
WWW'22.
paper / code / poster

Other activities

Reviewing activities

Neural Information Processing Systems (NeurIPS)'23-25
International Conference on Machine Learning (ICML)'24-25
International Conference on Learning Representation (ICLR)'24-25

Awards

3rd Place in Computer Science & Engineering, Samsung Humantech Paper, 2025.
2nd Place, Korea University Graduation Project, 2020, 2021.

Template based on Hojoon's website.