• Phone: +886 903 383 638
• Email: 63731417matthew@gmail.com
• Location: Taiwan
• GitHub: TyrantRey

Full-stack engineer specializing in RAG systems and artificial intelligence, with hands-on experience building production-grade systems from scratch. Independently developed multiple systems, including a semantic cache system that achieved 150× latency optimization and an AI phone customer service robot that compresses RAG-based text generation and audio generation into under 1 second. Published a conference paper on Multi-Hop RAG. Recipient of international student scholarships for three consecutive years.

• Programming Languages: Python, C, JavaScript, TypeScript, ShellScript

• Frontend: React, Next.js, Vue.js

• Backend: FastAPI, Flask

• Databases: PostgreSQL, MySQL, Milvus, Qdrant

• AI/ML: RAG, LLM Integration, Vector Search, YOLO

• Infrastructure: Docker, Proxmox, Grafana, Linux

Dec 2025 – present

Tech Stack: Python, Qdrant, FastAPI, Docker

• Designed company-wide LLM cache layer from scratch, serving as the unified entry point for all AI services

• Achieved 70% cache hit rate, reducing API costs by 20%

• Reduced response latency from 3 seconds to 20 milliseconds, achieving a 150× improvement

Oct 2025 – Dec 2025

Tech Stack: Python, C, SIP, WebSocket, STT/TTS, LLM, RAG

• Developed an AI voice robot supporting customer service calls and outbound marketing

• Implemented RTP packet transmission/reception modules in C, handling real-time audio stream encoding and decoding

• Built a SIP Gateway bridging to backend AI services via WebSocket

• Achieved end-to-end audio latency under 1 second, ensuring natural conversation experience

Sept 2025 – Oct 2025

Tech Stack: Grafana, Loki, Docker

• Designed and deployed a centralized logging system for multi-service environments

• Reduced log retrieval latency by 27% through indexing optimization

Sept 2024 – June 2025

Tech Stack: Python, FastAPI, PostgreSQL, Redis, LLM, Vector Database

• Led backend architecture design and LLM server deployment

• Developed AI-assisted search system, reducing patent search time from hours to 3 minutes, achieving a 90% reduction

• Optimized database queries, reducing search time by 20%

Dec 2025 – present

Tech Stack: Next.js, Docker

• Led development of AI learning assistance system, providing real-time semantic search based on academic literature

• Automatically generated questions using relevant academic content, adapting difficulty levels and course focus

• Integrated into university Moodle platform, enhancing student accessibility

Sept 2024 – Mar 2025

Tech Stack: Python, FastAPI, Milvus, Next.js, Docker

• Independently developed a customized RAG system supporting semantic search across thousands of academic documents

• Achieved 20% reduction in response latency through fine-tuning of embedding vectors

• Successfully integrated the system into the university's official website

Jan 2023 – Mar 2023

Tech Stack: Python, FastAPI, Vue.js, MySQL, Docker

• Independently designed and developed a full-stack project management system

• Implemented RBAC (Role-Based Access Control) for secure, role-specific access across multiple user types

Tech Stack: ShellScript, CI/CD

• Contributed open-source automation toolset, simplifying the deployment of LXC containers in Proxmox virtual environments

• Providing multiple one-click installation scripts for common services

Tech Stack: Translation, CI/CD

• Contributed to localization and continuous integration workflows

Tech Stack: Python, YOLOv11

• Developed a computer vision model by creating a custom dataset to identify various traditional Chinese medicinal herbs

Tech Stack: Python, Docker, MySQL

• Developed an asynchronous music streaming bot with real-time queue management

• International Student Scholarship Recipient for Three Consecutive Years (Academic Years 2023, 2024, 2025)

• Conference paper: Tracert-RAG: Boosting Multi-Hop RAG with Direction-Aware Graph Traversal (2025-08)