This page organizes my coursework and projects chronologically. It’s not just a list of assignments, but rather a reflection of my evolving capabilities, tracking my journey from “writing code” to “designing systems and researching problems.”

👉 Each item includes a GitHub repo; feel free to click through to read about topics that interest you.

🟢 Freshman Year — First Contact with Programming

Programming for Business Computing

🔖 Python OOP Pygame Game Development Creating an RPG game using Pygame was my first complete exposure to Python and object-oriented design. This was my first time transforming “writing code” into “making a complete product,” personally implementing everything from object-oriented design to game logic. This laid the foundation for my long-term use of Python as a core language. 📎 PBC2021-final

🔵 Sophomore Year — The Budding of NLP and System Implementation

Introduction to Text Mining

🔖 BERT NLP TensorFlow Flask Before ChatGPT emerged, I fully trained and deployed an sentiment classification system. This allowed me to truly understand the NLP pipeline (training → inference → deployment), marking a crucial starting point for my entry into the LLM / NLP field. 📎 toxic_comment_detector

Computer Programming (C++)

🔖 C++ Game Development Rendering Practiced C++ and low-level control capabilities, understanding performance and system behavior differences. Implemented a Windows OS-style game, transforming abstract logic into a real-time interactive system. 📎 windowskill_windowOS

Introduction to Python Data Analysis and Machine Learning

🔖 Machine Learning LangChain LLM Integration Integrated GPT-3 into a system to suggest how to rephrase emotionally manipulative sentences. This was my first attempt to connect “traditional NLP” with “early LLMs,” prompting me to consider how AI can practically change user behavior. 📎 toxic_comment_detector (LLM update)

Big Data and Business Analytics

🔖 Apriori Recommendation Systems Data Mining Designed a clothing recommendation system. This demonstrated my ability to translate business scenarios into data problems, even designing a functional recommendation logic without real transaction data. 📎 BDA2023_final_apriori

🟡 Junior Year — Databases, Cloud-Native, and Finance × CS

Database Management

🔖 DBMS SQL Normalization System Design Designed an airline data management system. Established the concept that “data structure design IS system design,” which profoundly influenced any subsequent projects requiring schema and data flow design. 📎 Airlines_DBMS

Cloud Native Application Development

🔖 React Express API Testing Stress Testing System Architecture A large interdisciplinary team project. Refactored parts of it in 2026, demonstrating my progress from mere participation to understanding the entire cloud-native architecture. 📎 CloudNative_Stadium_System

Robo-Advisor Special Topic Research

🔖 Quantitative Finance Risk Measurement Python Transformed financial background into a reusable Python modular analysis system, beginning to develop a mindset of “engineering financial problems.” 📎 robo_advisor

🟣 Senior Year — Information Retrieval and Exchange Period Exploration

Information Retrieval and Text Mining (IRTM)

🔖 RAG BM25 Vector Retrieval GraphRAG Systematically compared multiple RAG strategies in a Traditional Chinese context. Developed my research capabilities in IR (Information Retrieval) and LLM retrieval. 📎 IRTM_project

🔴 First Year Master’s — Deep Learning, LLM, and Research-Oriented

Deep Learning and Applications — Financial Time Series Research

🔖 GARCH LSTM TSMixer Loss Design Explored hybrid architectures combining traditional econometric models with deep learning. Attempted to reflect financial risk through loss design, representing a key highlight of my current research capabilities. 📎 GLoVE

Practical Computer Vision (AI Cup)

🔖 3D Medical Imaging Low-Resolution Learning Paper Reproduction Achieved 56th place out of 568 teams. Completed the full competition process from paper reproduction to model tuning. 📎 valve_abnormality_examination

Instruction Tuning — Classical Chinese ↔ Modern Chinese

🔖 QLoRA LLM Fine-tuning Prompt Engineering Fully completed the fine-tuning process using QLoRA, learning how to fine-tune LLMs under resource constraints and design inference strategies.

🧭 How to Read This Portfolio

You can explore in different ways:

  • By Topic: NLP / LLM, Finance × AI, Systems & Full-stack, Recommendation Systems
  • By Depth: Basic Implementation → Model Training → System Integration → Research-Oriented

✨ Conclusion

This timeline shows not just my projects, but a transformation in how I approach problems: from “implementing models” to “integrating them into systems,” and finally conducting research at the intersection of “Finance × AI.”