這個作品集導覽提供了一條清楚的閱讀路徑，而不單純是羅列所有專案。
我的背景橫跨財務分析、量化建模與系統導向的軟體開發，因此這裡的專案更重視「問題如何被拆解與重構」，而不只是功能或模型表現。
這份 portfolio 會持續滾動式更新。
每個被選入的專案，除了說明做了什麼，也會標註：
- 初次完成的時間與背景脈絡
- 後續 refactor 的時間點與動機
- 現在這個版本想解決的問題
因此閱讀這裡的方式，會比較像是在看一條隨時間演化的工程軌跡。
🔎 快速導覽（給不同背景的朋友）
不同領域的讀者可以從下表快速找到最相關的專案與技術內容。
| 如果你正在找… | 建議閱讀的專案 | 你會看到的能力與內容 |
|---|---|---|
| Finance × AI / Quant / FinTech | GLoVE robo_advisor Meme_stock_analysis_and_prediction | GARCH × 深度學習混合建模、模組化金融分析系統、另類資料與市場訊號研究 |
| LLM / RAG / NLP 系統開發 | 深度學習 HW3 – RAG system IRTM_project Toxic Comment Detector（BERT & GPT 版本） | Retriever / reranker 微調、繁體中文 RAG 策略實驗、LLM 出現前後的完整 NLP pipeline |
| LLM 微調 / Transformer 訓練 | Instruction tuning（QLoRA） Chinese Extractive QA（BERT） | 參數高效微調、span prediction 訓練流程、prompt 與推論策略設計 |
| Machine Learning / Deep Learning 基礎能力 | data_analysis AI Cup 3D 醫學影像建模與預測 | End-to-end 建模流程、時間序列預測、論文復現與實驗設計能力 |
| 系統整合 / 全端工程 | CloudNative_Stadium_System Airlines_DBMS | React + Express 架構、API 測試與壓力測試、資料庫 schema 與交易設計 |
| 推薦系統 / 資料探勘 | BDA2023_final_apriori | 關聯規則推薦系統與商業問題轉換能力 |
| 產品導向的 AI 應用 | youtube_nlp_analysis | 從使用者資料推導內容策略與商業洞察 |
| 程式設計基礎與系統思維起點 | PBC2021-final windowskill | OOP 設計、互動式系統實作、C++ 與 Python 的實作差異 |
🧭 建議閱讀路線
- GenAI / LLM 相關 → RAG system → IRTM → QLoRA
- Quant / FinTech 相關 → GLoVE → Robo-advisor
- 軟體工程 / 全端 → Cloud Native → DBMS
- ML Engineer → Data Analytics → AI Cup
💡 我目前的核心關注方向
- LLM / RAG 系統設計與最佳化
- Finance × AI 的可驗證建模
- 可落地的 Machine Learning 系統
建議閱讀專案
如果你是第一次來，建議從下列幾個專案開始，它們代表我在不同階段的思考方式：
1. Airlines_DBMS (資料庫設計與系統重構)
這是我將學術作業轉化為可操作系統的代表作。
- 2023.12 初版：學習 ER model 與正規化。
- 2026.02 重構：強化 schema 與應用層設計，讓使用者可實際操作查詢。
2. stock_analysis (金融量化工作流)
從簡單的分析導向轉向系統化的研究工作流。
- 2023.06 初版：基礎策略發想。
- 2026.03 重構：模組化資料處理，建立可重現的實驗結構。
最後
這份作品集也是我的工程筆記。如果你是為了面試或合作而來，謝謝你從這份導覽開始認識我。
This portfolio guide provides a clear reading path rather than simply listing all projects.
My background spans financial analysis, quantitative modeling, and systems-oriented software development. The projects here emphasize “how problems are decomposed and reframed” rather than just features or model performance.
This portfolio is continuously updated. Each selected project not only describes what was done, but also notes:
- When it was first completed and the context behind it
- When and why it was later refactored
- What problem the current version aims to solve
Reading this portfolio is more like viewing an engineering trajectory that evolves over time.
🔎 Quick Navigation (by Background)
Find the most relevant projects for your field from the table below.
| If you’re looking for… | Suggested Projects | Capabilities & Content |
|---|---|---|
| Finance × AI / Quant / FinTech | GLoVE robo_advisor Meme_stock_analysis_and_prediction | GARCH × deep learning hybrid modeling, modular financial analysis, alt-data |
| LLM / RAG / NLP Systems | Deep Learning HW3 – RAG system IRTM_project Toxic Comment Detector (BERT & GPT) | Retriever/reranker fine-tuning, Traditional Chinese RAG experiments |
| LLM Fine-tuning / Transformer Training | Instruction tuning (QLoRA) Chinese Extractive QA (BERT) | PEFT, span prediction training, prompt & inference strategy design |
| Machine Learning / Deep Learning Foundations | data_analysis AI Cup 3D Medical Imaging | End-to-end modeling, time series forecasting, paper reproduction |
| Systems / Full-stack Engineering | CloudNative_Stadium_System Airlines_DBMS | React + Express, API & load testing, database schema & transaction design |
| Recommendation / Data Mining | BDA2023_final_apriori | Association rule recommendations, translating business problems into models |
| Product-oriented AI Applications | youtube_nlp_analysis | Deriving content strategy and business insights from user data |
| Programming Foundations / Systems Thinking | PBC2021-final windowskill | OOP design, interactive systems, C++ vs Python implementation differences |
🧭 Suggested Reading Paths
- GenAI / LLM → RAG system → IRTM → QLoRA
- Quant / FinTech → GLoVE → Robo-advisor
- Software Engineering / Full-stack → Cloud Native → DBMS
- ML Engineer → Data Analytics → AI Cup
👉 View the Complete Project Timeline
💡 Current Focus Areas
- LLM / RAG system design & optimization
- Verifiable modeling for Finance × AI
- Deployable Machine Learning systems
Suggested Starting Projects
If this is your first visit, I recommend starting with these projects — they represent my thinking at different stages:
1. Airlines_DBMS (Database Design & System Refactoring)
A showcase of how I transformed an academic assignment into a fully operable system.
- Dec 2023 (Original): Learning ER modeling and normalization.
- Feb 2026 (Refactored): Enhanced schema and application layer for real user queries.
2. stock_analysis (Financial Quantitative Workflow)
From simple analysis toward a systematic research workflow.
- Jun 2023 (Original): Basic strategy ideation.
- Mar 2026 (Refactored): Modular data processing with a reproducible experiment structure.
Finally
This portfolio is also my engineering notebook. If you’re here for an interview or collaboration, thank you for starting with this guide.