cc-switch
Resolved cross-platform MCP compatibility issues across Gemini CLI timeout handling, WSL path detection, and Windows test failures.
About Me
- Computer Science graduate with solid algorithmic training and graduate-level AI/ML coursework.
- I build Python and Rust services, LLM integrations, MCP tools, and search-backed applications.
- Open-source contributor to high-visibility projects, with merged PRs improving API compatibility, cross-platform behavior, and production reliability.
- I also maintain Discord bot tooling for scheduled coding challenges, submission tracking, and LLM-assisted workflows.
Open Source Contributions
gemini-balance
Implemented Gemini Files API and native embedding endpoint support across models, service layer, authentication, persistence, error handling, and logging.
Instaloader
Migrated Instagram profile metadata fetching from deprecated endpoints to GraphQL while preserving existing API compatibility.
Education
National Yang Ming Chiao Tung University
Graduate Coursework in Computer Science and Engineering
2024 - present
Focus: Machine Learning and Computer Vision.
National Cheng Kung University
B.S. in Computer Science and Information Engineering
2017 - 2023
Focus: NLP and Information Retrieval.
Skills
Programming
Python
Rust
C++
JavaScript
C
Verilog
Dart
Backend
FastAPI
Flask
axum
REST APIs
GraphQL
SQLite
AI / LLM Tooling
LLM API Integration
MCP
Embeddings
Vector Search
Agentic Coding Tools
Developer Tools
Git
Docker
GitHub Actions
CI/CD
WSL
CLI Tooling
Side Projects
GeminiASR
CLI transcription pipeline for long-form audio/video with chunking, parallel Gemini API requests, key rotation, batch processing, SRT export, and TOML configuration.
oj-api-rs
Backend for competitive-programming problem retrieval with crawler workflows, SQLite persistence, admin dashboard, OpenAPI docs, MCP integration, and vector-similarity search.
leetcode-daily-discord-bot
Discord bot used across dozens of servers for LeetCode daily challenges, timezone-aware scheduling, slash commands, submission tracking, i18n, LLM integration, and caching.
Competitive Programming
2024 Collegiate Programming Examination: 6/7, ranked 51 / 2668 (Top 1.9%).
Selected Coursework
Deep Learning & Deep Learning Labs
Read original papers, then implemented and trained SCCNet, ResNet, UNet, Conditional VAE, MaskGIT, and DDPM. The final project explored scheduler variants to improve RePaint computation efficiency.
Data Visualization and Visual Analytics
Visualized Taiwan regional voting results and analyzed relationships between voting patterns, income, and education levels.