1. General Purpose

To create a digital presence that reflects my engineering philosophy: using software as a tool to solve structural problems. Rather than using basic HTML/CSS or pre-built website builders, the goal was to master the modern Next.js App Router architecture. It aims to showcase not just my past projects, but my ability to design scalable component trees, manage complex dynamic routing for a markdown-based blog, and implement clean, maintainable styling systems.

2. My Role

Full-stack Developer and UI/UX Architect. I engineered the entire frontend architecture, ensuring strict type safety with TypeScript to catch errors during development. I implemented a dynamic markdown rendering system with full KaTeX/LaTeX support for mathematical formulas, optimized asset loading for maximum performance, and crafted a minimalist interface that prioritizes readability and seamless user experience.

3. Project Link

1. General Purpose

To build a fully functional, high-performance multiplayer mobile game from scratch, handling the complex engineering challenge of synchronizing real-time data from 8 different players with millisecond latency. The core objective was to master custom UDP networking, complex algorithmic logic for collision and penalty pools, and backend containerization. Beyond the game itself, the purpose was to use this product as a real-world sandbox to simulate a tech startup—learning how to create rigorous business plans and understanding the life cycle of a product from a commercial perspective.

2. My Role

System Architect & Simulated CTO. On the technical side, I engineered the custom UDP sockets for real-time synchronization, designed a server-authoritative backend to prevent client-side manipulation, and fully containerized the server environment using Docker for professional, isolated deployment. On the management side, I ran a meticulous operational simulation of a 'Shadow Company.' I calculated hypothetical employee salaries, estimated development timelines based on strict man-hours, and conducted rigorous cost-benefit analyses for infrastructure. For example, I mathematically weighed the financial impact of purchasing a commercial Firewall license versus paying developers' salaries to build an internal security solution.

3. Project Link

1. General Purpose

The primary objective is to solve the critical trade-off between speed and stability in machine learning optimization. While Nesterov Accelerated Gradient (NAG) provides rapid convergence, its momentum mechanism causes severe instability and overshooting in high-noise mini-batch environments. Conversely, Stochastic Gradient Descent (SGD) is highly stable but notoriously slow. This project aims to mathematically formulate a dynamic transition rule that bridges these two methods, monitoring cost oscillation to optimize neural network training.

2. My Role

Project Lead and Core Researcher. I formulated the theoretical mathematical foundation of the AGGOA algorithm, defining the dynamic transition coefficient based on gradient magnitude and cost oscillation. I then transitioned this theory into practice by prototyping the algorithm in Python using PyTorch, and conducting rigorous comparative experiments (AGGOA vs. Adam vs. NAG) on Convolutional Neural Networks (CNNs) using the CIFAR-10 dataset to quantitatively prove improvements in convergence speed and divergence mitigation.

1. General Purpose

To build a performant mobile application that maintains a unified codebase while interacting efficiently with RESTful APIs. It served as a deep dive into asynchronous data fetching and mobile UI/UX paradigms. More importantly, the project's cancellation highlighted the non-technical realities of software engineering: the necessity of continuous stakeholder alignment and adapting to shifting market demands.

2. My Role

Lead Mobile Architect. I designed the component hierarchy using React Native and NativeWind, handled complex API calls, and managed application state. Despite the technical achievements, the project was ultimately halted because the target audience demand decreased and the project partners could not maintain coordination—teaching me that engineering success requires both solid code and strong business alignment.