{{ lang === 'en' ? "Master's Student in Data Science and Engineering" : '數據科學與工程 碩士生' }}
{{ lang === 'en' ? 'Machine Learning, Deep Learning, Data Science Project, Biostatistics' : '機器學習、深度學習、資料科學專題、生物統計' }}
{{ lang === 'en' ? 'Probability and Statistics, Fourier Analysis, Data Structures and Algorithms' : '機率與統計、傅立葉分析、資料結構與演算法' }}
{{ lang === 'en' ? 'Enhanced Video Pose Transformer (VPT) by integrating Mamba-based sequence blocks for GPU-accelerated 2D-to-3D human pose estimation. Reduced inference latency and training cost through CUDA-level optimization, parallelized data flow, and modular system integration.' : '強化現有之 VPT 模組，整合 Mamba 序列區塊以提升 GPU 加速的 2D 至 3D 人體姿態估計效率。透過 CUDA 層級優化、平行化資料流及模組化系統整合，降低推論延遲與訓練成本。' }}
{{ lang === 'en' ? 'Data conversion and algorithm development for data collection systems.' : '協助院方數據收集裝置進行資料轉換與演算法開發。' }}
{{ lang === 'en' ? 'Tailored teaching with real-life examples, maintaining progress tracking with students and parents.' : '依學生程度與吸收狀況，結合生活實例協助理解課程，並與學生和家長保持互動，共同掌握進度與設定目標。' }}
{{ lang === 'en' ? 'Designed and implemented an AI-powered agent system for automated code refactoring. Built multi-step reasoning pipelines that analyze code structure, detect anti-patterns and code smells, and apply targeted transformations to improve maintainability, readability, and performance while preserving functional correctness through automated test validation.' : '設計並實作 AI 驅動的代理系統進行自動化代碼重構。建構多步驟推理管線，分析程式碼結構、偵測反模式與程式碼異味，並應用針對性轉換以提升可維護性、可讀性與效能，同時透過自動化測試驗證確保功能正確性。' }}
{{ lang === 'en' ? 'Fine-tuned Llava model with LoRA, applied data augmentation and image filtering. Researched Flamingo and Honeybee as alternatives.' : '利用 LoRA 微調 Llava 基礎模型，應用資料增強技術提升模型泛化能力。' }}
{{ lang === 'en' ? 'Trained ensemble models (XGBoost, Random Forest) and TabNet for stock prediction with correlation-based feature selection.' : '以 AI 技術預測台灣股市中的指數增長型股票，訓練 XGBoost、Random Forest 及 TabNet 模型。' }}
{{ lang === 'en' ? 'Formulated graph-based job-shop optimization using Integer Linear Programming (ILP) for optimal scheduling. Explored Greedy, DP, RL, and Genetic Algorithm approaches.' : '自動化系統 Scheduling 和 AI Agent 創意提案，運用整數線性規劃求解最佳排程。' }}
{{ lang === 'en' ? 'Processed and augmented training data with image normalization and mask-conditioned augmentation. Evaluated with FID.' : '以生成式 AI 建構無人機於自然環境偵察時所需之導航資訊，影像資料生成競賽。' }}
{{ lang === 'en' ? 'Ensembled LightGBM and XGBoost for regression. Cleaned and merged multi-source datasets with hyperparameter tuning.' : '清理並合併多來源資料集，融合 LightGBM 和 XGBoost 提升迴歸準確度。' }}
{{ lang === 'en' ? 'LoRA + Adapter achieved +3.6% accuracy on vision tasks (0.8775 → 0.9134) with only 1.6% trainable parameters. LoRA + BitFit improved NLP accuracy on RoBERTa-base (0.931 → 0.945 on SST-2).' : '結合 LoRA 與 Adapter 在視覺任務上提升 3.6% 準確度，僅使用 1.6% 可訓練參數。LoRA + BitFit 在 RoBERTa-base 上提升 NLP 準確度。' }}
{{ lang === 'en' ? 'Built classification pipelines with stratified k-fold CV and grid-search tuning. Achieved Accuracy 83.6%, ROC-AUC 91.2%, F1 86.3%.' : '建立分類管線，使用分層 k-fold 交叉驗證與網格搜尋調參，達到準確度 83.6%、ROC-AUC 91.2%。' }}
{{ lang === 'en' ? 'Built real-time ocean scene in C++/OpenGL with dynamic wave animation, DuDv refraction, cubemap skybox, and Fresnel reflectance.' : '以 C++/OpenGL 建構即時海洋場景，包含動態波浪動畫、DuDv 折射、天空盒與菲涅耳反射。' }}
{{ lang === 'en' ? 'Analyzed professional match data and framed checkout/target selection as a game-theoretic decision process.' : '分析職業比賽數據，將結帳/目標選擇建構為賽局理論決策過程。' }}
{{ lang === 'en' ? '2020 National Collegiate Darts Championship · 9th Asian International Darts Open — Doubles B Division 5th Place' : '2020 全國大專飛鏢錦標賽 · 第九屆亞洲國際飛鏢公開賽 雙人組 B 級第五名' }}