cv

My curriculum vitae showcasing my research experience, education, awards, and technical skills in AI and deep learning.

Basics

Name Yejun Jang
Label AI Researcher
Email jangyejun@snu.ac.kr
Url https://mutantq.github.io
Summary AI researcher and pre-founder of mutual, building trust infrastructure for the post-AI media era. Research focus on developing reinforcement learning agents that can make reliable decisions based on grounded factual information, bridging advanced decision-making algorithms with trustworthy information infrastructure. Currently pursuing B.S. in Electrical and Computer Engineering at Seoul National University.

Interests

Artificial Intelligence
Multi-agent Systems
Reinforcement Learning
Deep Learning
AI Security
Applied AI
Robotics
Penetration Testing

Publications

Work

  • 2025.10 - Present

    Seoul, South Korea

    Pre-founder
    mutual
    Building trust infrastructure for the post-AI media era through hardware-rooted content authenticity.
    • Designing SRA (Signing Right Away), a hardware-rooted architecture for C2PA-aligned content authenticity
    • Integrating authenticated encryption and secure camera pipelines between image sensors and trusted execution environments
    • Raised angel investment from Kay Kyungsik Woo (MVL Foundation) to support FPGA prototyping and hardware development
  • 2024.04 - 2025.10

    Daejeon, South Korea

    Military Science & Technology Researcher
    Republic of Korea Army, Innovation Institute for Future Army (IIFA)
    Mandatory military service as a Military Science & Technology Researcher, focusing on AI security and deep reinforcement learning applications for military systems.
    • Conducted research on AI security, focusing on model and data poisoning attacks on military AI systems
    • Investigated deep reinforcement learning applications for penetration testing on large-scale networks
    • Presented research findings at national conferences (KICS 2025, CISC-W 2024)
    • Led work that resulted in publications at KICS 2025 and CISC-W 2024 on DQN-based penetration testing and supply chain verification for military AI systems
  • 2023.12 - 2024.03

    Daejeon, South Korea

    Research Intern
    KAIST ACSS Lab
    Advisor: Prof. Soojean Han.
    • Explored applications of group equivariant neural networks for multi-agent traffic optimization
    • Investigated traffic light control systems using multi-agent reinforcement learning
    • Gained experience in symmetry-aware neural network architectures and their applications to real-world problems
  • 2023.02 - 2024.02

    Seoul, South Korea

    Project Head, SOUL - A Modern Reinterpretation of Keynes' Bancor
    Korea Foundation for Advanced Studies (KFAS)
    Modeled international trade using multi-agent reinforcement learning to simulate Keynes' International Clearing Union.
    • Modeled central banks of each country as an RL agent aiming to maximize annual GDP in a simulated global economy
    • Explored economic policy and international relations through game-theoretic and multi-agent simulations
    • Secured funding of 10 million KRW (~$10,000 CAD) and managed allocation for computing, mentoring, and resources
  • 2022.08 - 2023.11

    Seoul, South Korea

    Editor, Questions for Future
    KFAS & Institute for Future Strategy, Seoul National University
    Branstormed impactful keyphrases capturing the essence of the digital society.
    • Coined the phrase “attention ownership” – the right to control, manage, and determine how one's attention is directed.
    • Book published with title 'Living Ideas from Future Observers (미래 관찰자의 살아있는 아이디어)', available for purchase throughout Korean bookstores.
  • 2019.02 - 2020.02

    Sejong, South Korea

    Software Lead, FHP-RTFS
    Sejong Academy of Science and Arts (SASA)
    Built a contactless feedback system for alerting incorrect body posture when using digital devices.
    • Developed a measurement criterion for Forward Head Posture (FHP) using computer vision techniques
    • Processing 3D webcam data and providing audio feedback led to 81% improvement in participants' FHP
    • Became 2020 Regeneron ISEF finalist

Education

  • 2021.03 - 2026.12

    Seoul, South Korea

    B.S.
    Seoul National University
    Electrical and Computer Engineering
    • Research Focus: Deep Reinforcement Learning, Generative Models, AI Safety
    • Expected graduation December 2026
  • 2018.03 - 2021.01

    Sejong, South Korea

    High School Diploma
    Sejong Academy of Science and Arts (SASA)
    Computer Science and Physics

Awards

Skills

Deep Learning & AI
Reinforcement Learning: Offline RL, Multi-agent RL, Deep Q-Learning, Policy Gradient Methods
Generative Models: Flow Matching, Normalizing Flows, VAE, GAN
Deep Learning: Sequential Models (LSTM, Transformers), Function Approximation
AI Security: Adversarial Attacks, Model Poisoning, Supply Chain Security
Programming Languages
Python
C
C#
Bash
Libraries & Frameworks
PyTorch, JAX, huggingface, NumPy, Pandas, Matplotlib, Docker, Git, uv
RL: PettingZoo, Gymnasium (OpenAI Gym), RLlib, CleanRL, trl
Research & Communication
Technical Writing, Academic Research, Conference Presentations
Cross-disciplinary Communication (AI, Economics, Philosophy)

Languages

Korean
Native speaker
English
Fluent

Projects

  • 2024.03 - Present
    Signing Right Away (SRA)
    Hardware-rooted trust architecture for verifiable digital provenance and content authenticity, forming the core of mutual's SRA (Signing Right Away) system.
    • Designed hardware-rooted trust architecture for verifiable digital provenance and C2PA-aligned content authenticity
    • Developed authenticated encryption scheme (e.g., AES-GCM/CCM) for securing camera-to-processor image data
    • Proposed TEE-based immediate signing to combat AI-generated fake content
    • Raised angel investment to fund FPGA-based prototyping and secure camera pipeline development
  • 2021.08 - 2021.08
    QuPid - VR Quantum Computing Education
    Created a VR laboratory to explain basic but hard-to-grasp concepts in quantum computing, specifically the qubit.
    • Virtual Reality
    • Educational Technology
    • Quantum Computing
  • 2021.09 - 2021.12
    Car-The-Garden - Autonomous Navigation
    Applied left-first and right-first search algorithms to navigate a miniature version of the standard Korean driving test.
    • Autonomous Navigation
    • Search Algorithms
    • Robotics

Volunteer

  •
    Volunteer Educator
    South Vancouver Neighborhood House (SVNH)
    Assist community engagement programs and teach math and science to community members.
  •
    Translator
    Raspberry Pi Foundation
    Translated multiple educational articles in the Raspberry Pi Foundation's Project page to Korean.