柯宥圻 YU-CHI (Yuchi) KO

✉ r14946005@ntu.edu.tw ｜ GitHub ｜ LinkedIn

AI 系統與 LLMOps｜Agentic AI｜軟體工程

🎓 學歷 EDUCATION

國立臺灣大學 National Taiwan University

資料科學碩士 M.S. in Data Science (College of EECS) Taipei, Taiwan | Sep 2025 – Present

  • 實驗室: Machine Intelligence & Agentic System Lab (MSLAB)
  • 指導教授: 林守德 教授 Prof. Shou-de Lin
  • 研究方向: LLM 應用、智慧交易系統、Agent 系統

國立臺灣大學 National Taiwan University

財務金融學學士 ＆ 決策資料科學雙主修 B.B.A. in Finance & Decisive Data Science (UIBP) Taipei, Taiwan | Sep 2021 – Jun 2025

  • GPA: 4.04 / 4.3 (Total 192 credits)

💼 經歷 EXPERIENCE

國泰世華銀行 Cathay United Bank

LLMOps 工程師 LLMOps Engineer Taipei, Taiwan | Jun 2024 – Dec 2024

  • 企業級 AI 系統落地：共同開發並上線國泰首套內部法規查詢 AI 系統。顯著降低 hallucination、提升金融領域正確性，檢索準確率由 70% 提升至 94%。
  • 進階 RAG 與 GraphRAG 架構：建立金融與法遵文件專用前處理流程，導入 entity linking 與 multi-hop reasoning。
  • Agent 工作流整合：導入公司首個正式上線的 agent-based LLM 系統，支援 memory、tool use、圖表與簡報生成。建立具資安意識的惡意查詢偵測模組，安全拒答率提升 85%。

沛智科技 Genibuilder Inc.

AI 工程實習生 AI Engineer Intern Taipei, Taiwan | Jul 2025 – Aug 2025

  • 啟發式 Agent 推理框架：打造 AlphaEvolve 啟發式架構（GA + LLM reasoning），在 100 節點 TSP 達到 4.8% gap，token 使用量降低約 20 倍。
  • 多雲部署：於 GCP 與 Azure 部署容器化 LLM 服務。

中央研究院 Academia Sinica (IIS)

大學部研究助理 Undergraduate Research Assistant (NLP) Taipei, Taiwan | Jan 2024 – Jun 2024

  • 財報重點擷取資料集：建立人工標註之財報資料集，用於決策導向 LLM 評測。

豹投資 Gorich (Financial AI Startup)

Prompt Engineer / 軟體工程師 Taipei, Taiwan | Aug 2023 – Feb 2024

  • 早期 LLM 落地 (2023)：在企業級工具成熟前，即建置 LangChain RAG 系統並實際上線。台灣最早期的 production-grade RAG 系統之一。

🚀 專案 PROJECTS

  • GraphRAG 與檢索消融研究：分析金融與醫療文件的準確率、延遲與成本權衡。
  • 55688 使用者偏好預測：建立使用者意圖分類系統，較 baseline 提升 11%。
  • LSTM + GARCH 波動度預測：設計混合式時序模型，在市場轉折 (regime shift) 下展現較高穩定性。

🏆 其他榮譽與技能 ADDITIONAL

  • 競賽相關：
    • GenAI STAR 全國競賽：前 10 名 (Top 10 / 700+ teams), ELTV 組最佳獎。
    • 2026 TSMC IT CareerHack：決賽入圍 (Finalist).
  • 語言能力：中文 (Native), English (IELTS 7.5; TOEIC 895), Spanish (Intermediate).
  • 國際交流：巴塞隆納 6 個月交換學程。
  • 技術社群：台大 AI 社 技術長 (NTUAI Technical Officer).

🛠 技術能力 TECHNICAL SKILLS

  • 核心技術: Python, C++, SQL, Docker, Git, Linux, FastAPI, PostgreSQL, Redis, Neo4j, Cloud (GCP, Azure).
  • LLM 系統: RAG, GraphRAG, LLM Agents, Tool Use & Memory, Fine-tuning, LangChain, LlamaIndex.

YU-CHI (Yuchi) KO

✉ r14946005@ntu.edu.tw ｜ GitHub ｜ LinkedIn

AI Systems & LLMOps · Agentic AI · Software Engineer

🎓 EDUCATION

National Taiwan University

M.S. in Data Science (College of EECS) Taipei, Taiwan | Sep 2025 – Present

  • Lab: Machine Intelligence & Agentic System Lab (MSLAB)
  • Advisor: Prof. Shou-de Lin
  • Research Focus: LLM applications, intelligent trading systems, and agent systems

National Taiwan University

B.B.A. in Finance & Decisive Data Science (UIBP) Taipei, Taiwan | Sep 2021 – Jun 2025

  • GPA: 4.04 / 4.3 (Total 192 credits)

💼 EXPERIENCE

Cathay United Bank

LLMOps Engineer Taipei, Taiwan | Jun 2024 – Dec 2024

  • Production-Scale AI System: Co-developed and launched Cathay’s first internal AI system for regulatory queries; improved finance-domain correctness and raised retrieval accuracy from 70% to 94%.
  • Advanced RAG & GraphRAG Pipeline: Implemented enterprise-grade RAG with preprocessing for financial & compliance documents; added entity-linking and multi-hop reasoning.
  • Agent-based Workflow Integration: Introduced multi-agent orchestration — the company’s first production-ready agent system supporting memory, tool use, chart and slide generation. Built a security-aware malicious query detection module, improving safe refusal rate by 85%.

Genibuilder Inc.

AI Engineer Intern Taipei, Taiwan | Jul 2025 – Aug 2025

  • Genetic Search + Agentic Optimization: Built an AlphaEvolve-inspired agent framework combining GA + LLM reasoning; achieved a 4.8% gap on 100-node TSP while reducing token usage by ~20x.
  • Multi-Cloud Deployment: Deployed containerized LLM services across GCP and Azure.

Academia Sinica (IIS)

Undergraduate Research Assistant (NLP) Taipei, Taiwan | Jan 2024 – Jun 2024

  • Financial Report Highlighting Dataset: Constructed a human-annotated financial report dataset for decision-oriented LLM evaluation.

Gorich (Financial AI Startup)

Prompt Engineer / Software Engineer Taipei, Taiwan | Aug 2023 – Feb 2024

  • Early LLM Adoption (2023): Entered the LLM field extremely early (mid-2023) and built one of Taiwan’s first production-grade LangChain RAG systems before enterprise tooling matured.

🚀 PROJECTS

  • GraphRAG & Retriever/Reranker Ablation Study: Evaluated retrievers and rerankers for financial and medical documents on accuracy, latency, and cost trade-offs.
  • User Preference Prediction: Built a multi-category user intent classification system, outperforming baseline by 11%.
  • LSTM + GARCH Volatility Forecasting: Designed hybrid LSTM+GARCH volatility models with improved stability under market regime shifts.

🏆 ADDITIONAL

  • Awards:
    • GenAI STAR National Competition: Top 10 / 700+ teams, Best Award in ELTV category.
    • 2026 TSMC IT CareerHack: Finalist.
  • Languages: Mandarin (Native), English (IELTS 7.5; TOEIC 895), Spanish (Intermediate).
  • International Exchange: 6-month exchange program in Barcelona.
  • Technical Leadership: NTU AI Club Technical Officer (NTUAI).

🛠 TECHNICAL SKILLS

  • Core: Python, C++, SQL, Docker, Git, Linux, FastAPI, PostgreSQL, Redis, Neo4j, Cloud (GCP, Azure).
  • LLM Systems: RAG, GraphRAG, LLM Agents, Tool Use & Memory, Fine-tuning, LangChain, LlamaIndex.