👋 哈囉，我是 Scott（scott0127）

我目前就讀於 國立中正大學資訊工程研究所（CCU CSIE），是一名資訊工程碩士研究生。
我的開發經驗涵蓋 全端開發、AI / LLM 應用、GraphRAG / RAG、嵌入式系統、DevOps 與雲端部署。

我喜歡從真實使用者需求出發，透過軟體工程與 AI 技術解決具體問題。
目前主要關注 LLM Workflow、GraphRAG、全端系統架構、Edge AI、Kubernetes / Docker 部署 等方向。

---

🚀 About Me

• 🎓 M.S. student in Computer Science and Information Engineering at National Chung Cheng University
• 🏆 1st Place, 2024 National University Smart Innovation and Interdisciplinary Integration Competition
• 🧠 Interested in LLM, GraphRAG, RAG, Agentic Workflow, Computer Vision, and Full-stack Development
• 🛠️ Experienced in building systems from frontend, backend, database, deployment, to edge devices
• 🌏 Built and operated a real-world web tool with thousands of registered users
• 📚 Passionate about turning technical ideas into useful, user-centered applications

---

🛠️ Languages and Tools

Languages

C++



Python



TypeScript

Frontend & Desktop

Vue.js



Nuxt.js



Tailwind



Electron

Backend & Database

Node.js



NestJS



FastAPI



PostgreSQL



MySQL



Redis

AI / Machine Learning / Edge AI

PyTorch



TensorFlow



OpenCV



Jetson

DevOps & Platforms

Docker



K8S



GCP



Linux



Git



GitHub

---

🔬 Technical Focus

LLM / RAG / Agentic Workflow

• Built LLM workflows for long-form lecture content processing
• Worked with GraphRAG / RAG to improve knowledge retrieval and reduce hallucination
• Designed pipeline-style AI workflows for summarization, chapter segmentation, and educational assistant systems
• Interested in Multi-Agent Systems, prompt engineering, and AI-powered learning tools

Full-stack Development

• Frontend: Nuxt 3, Vue.js, TypeScript, Tailwind CSS
• Backend: FastAPI, NestJS, Node.js
• Database: PostgreSQL, MySQL, Redis, Supabase
• Deployment: Docker, Kubernetes, Nginx, GCP, Render

Edge AI / Embedded Systems

• Experience with NVIDIA Jetson Nano
• Worked on computer vision, image processing, object detection, OCR, and edge deployment
• Used YOLOv8, PaddleOCR, ONNX / TensorRT-related deployment concepts

---

📌 Featured Projects

🧠 IntelliTeach — AI-integrated Smart Learning Platform

IntelliTeach is a smart learning platform integrating Edge AI, LLM, GraphRAG, and embedded classroom devices.
The system solves the problem of preserving physical blackboard content and helping students review long lecture videos more efficiently.

Key features:

• Built a distributed system combining frontend, backend, cloud GPU resources, and embedded devices
• Used NVIDIA Jetson Nano for edge-side computer vision processing
• Implemented blackboard image processing, silhouette removal, and perspective transformation
• Designed an LLM workflow for long lecture transcripts and course material processing
• Applied GraphRAG to build structured course knowledge and power an AI teaching assistant
• Containerized services using Docker and deployed with Kubernetes
• Won 1st Place in the 2024 National University Smart Innovation and Interdisciplinary Integration Competition

Tech stack:

Nuxt 3 FastAPI Python Docker Kubernetes GraphRAG LLM NVIDIA Jetson Nano Computer Vision

---

🌱 Pikmin Bloom Collection Tracker / Map Tool

A web tool built for the Pikmin Bloom player community, focused on decoration collection tracking and map-based discovery.

Key features:

• Built with Nuxt 3 and backend services
• Integrated map data and filtering logic for decoration discovery
• Used service workers and data-splitting strategies to reduce server output traffic
• Improved SEO and achieved strong Google Search visibility for related keywords
• Operated as a real-world user-facing tool with thousands of registered users

Tech stack:

Nuxt 3 Vue.js TypeScript Supabase PostgreSQL Service Worker SEO Map Data

---

🚦 WalkerGuard — Pedestrian Safety Violation Detection System

WalkerGuard is an embedded AI system designed to improve pedestrian safety by detecting vehicle violations near intersections.

Key features:

• Used YOLOv8 for vehicle detection
• Applied OCR for license plate recognition
• Designed violation detection logic based on traffic rules
• Deployed on NVIDIA Jetson Nano for edge computing
• Worked with multi-view tracking and high-speed plate capture concepts

Tech stack:

Python YOLOv8 PaddleOCR NVIDIA Jetson Nano Computer Vision Edge AI

---

🏛️ HistoSphere — Multi-Agent Historical Learning System

HistoSphere is an AI-powered interactive learning system that allows users to explore historical events through conversations with AI-generated historical figures.

Key features:

• Designed a Multi-Agent System for historical character responses
• Built workflow pipelines for historical data collection, crawling, and RAG
• Used multimodal generation to create historical scenes with consistent visual style
• Integrated educational psychology concepts such as scaffolding and cognitive load reduction

Tech stack:

Nuxt 3 Python Docker Nginx RAG Multi-Agent System LLM Workflow Playwright

---

🏆 Achievements

• 🥇 1st Place, 2024 National University Smart Innovation and Interdisciplinary Integration Competition
• 🎓 Received academic excellence awards during undergraduate study
• 📈 Built and operated a real-world web service with strong SEO visibility and thousands of users
• 👨‍🏫 Experience teaching C++, AI fundamentals, and Python programming

---

📊 GitHub Insights & Open Source Impact

📈 GitHub Overview

---

🌟 Open Source Impact

• ⭐ Earned GitHub stars across personal projects, showing real-world project visibility and usefulness
• 🔀 Contributed pull requests to external open-source projects
• 🤝 Became a contributor to an open-source project with 1.1k+ GitHub stars
• 🧩 Experienced in collaborating with existing codebases, following project conventions, and contributing production-ready changes

---

📉 Contribution Activity

📫 Contact

Collapse