Hanbie Ryu - Profile

I was born in Sheffield, England and I am currently majoring in Artificial Intelligence at Yonsei University

I have over 7 years of experience in C# development using Unity, and I am proficient in Python for machine learning and data analysis. Combining these areas, I aim to research better ways to develop more intelligent systems in virtual environments like video games, or applying them to robotics to solve complex real-world problems.

Deep reinforcement learning and applying advanced AI agents to
virtual and physical environments to tackle complex, challenging tasks

Education

- Yonsei University, Dept. of Artificial Intelligence (B.S, Sophomore) March 2023 – Present

- Cumulative GPA: Overall: 4.06/4.3 | Major: 4.15/4.3 | (Ranked 4/47 in class)

Coursework: Computer Programming, Object-Oriented Programming, Data Structures, Discrete Mathematics,
Linear Algebra, Probability and Statistics, CS231n (Stanford University open course)

- Project Highlights -

Evolution Sim in Unity - 2024

1 | Graph-based NeuroEvolution of Augmented Topologies (NEAT) simulation

2 | Neural network system with mutations and genetic crossovers

3 | Live visualizer to enable real-time analysis

4 | Includes various environments:

- Ecosystem with agents that search for food to survive and reproduce
- 2-team large-scale shooter game to simulate competitive evolution
- 2D Drone simulation for learning complex physical controls via evolution

[github.com/HBRYU/ShittyDarwinInc]

Blade Weaver - 2023 ~ 2024

1 | Game development passion project, made in Unity (C#)

2 | Fluid combat controls, advanced AIs and procedural animations

3 | Pixel graphics and animations are made in Aseprite, SFX in FL Studio

4 | Talk at 2023 SKYST Conference, Seoul Campus, KAIST, August 2023

5 | [github.com/HBRYU/Tower_B-40]

"Blade Weaver", aka "Tower B-40" is a game I first came up with in 2017 with a few iterations over the years.
I began its current development in 2023, and it is still ongoing as a side project, planned for release in the future.

- Optimized pathfinding over long distances by combining graph-based and grid-based A* pathfinding algorithms, improving performance and precision
- Live pixel-perfect rendering to align on-screen elements to world grid while maintaining classic pixel art aesthetic
- Mesh-based "wings" with precise, live vertex calculations for fluid and dynamic visuals in combat
- Enemy AI with state machine architecture that predicts player movement and calculates target positions for each limb for optimal movement in physical environment
- Live inverse-kinematics calculations
- Background parallaxing and smooth camera controls for optimal player experience
- Mechanical SFXs designed using frequency modulation synthesis
- Audio pooling for fast, on-demand audio source generation

- Research Experience -

CT Image Denoising at Medisys Lab, Yonsei University - 2024

As an undergraduate research intern, I understood and implemented image processing solutions using a variety of CNN architectures and training methods

Personal contributions:

- a) Experimented with REDCNN and U-net architectures for CT denoising
- b) Developed a codebase for managing different model architectures and experiments
- c) Implemented a WGAN-GP system to prevent over-smoothing of generated CT images

[Medisys Lab]

Technologies

1 | Languages: C, C++, C#, Python, HTML, Javascript
2 | Machine Learning: PyTorch, TensorFlow, Keras
3 | JS Frameworks: React + Expo, NodeJS
4 | Collaborative Tools: Git, Docker, Tmux
5 | Embedded Systems: Arduino, RaspberryPi
6 | Tools: Unity, Photoshop, Blender, FL Studio, Audacity, Premiere Pro

Hanbie Ryu | HBRYU

Engineer, Designer, & Game Developer

- Research Interests -