//assets.cakeresume.com/assets/contentarea/contentarea-2d64eca8dae2a1ffb31e0fb1df887bbad072c7c19734d7d30fa14884bbcc299a.css

TOEIC - 855

The TOEIC Program

預計六七月畢業，正在找全職工作！

國立臺北科技大學(National Taipei University of Technology, Taipei Tech)

淡江大學（Tamkang University）

Software Engineer Intern @ LINE

Yu-Tung Chen (陳禹彤)

LINE Taiwan Limited

軟體工程實習生

快畢業再不找工作就要睡公園的鹹魚軟體工程師

· 參與內部生成式人工智慧研究討論並給予一些關於 LLM 和 Stable Diffusion 的建議。 
· 在 40 多人面前分享了關於如何成為軟體工程師的演講。 
· 利用 Jenkins 和 GitHub Actions 等 CI 工具為所有 scrum 團隊部署自動化測試案例分類系統，包括定期備份、資料預處理和機器學習分類。 
· 增強整體可用性和開發人員體驗，例如實施自動化測試、自動發布最新產品版本以及將過時的框架遷移到重大變更版本。

· Participated in internal Generative AI research discussions and gave some advice about LLM and Stable Diffusion.
· Shared a speech about how to become a software engineer in front of more than 40 people.
· Leveraged CI tools such as Jenkins and GitHub Actions to deploy automated test cases classification system for all scrum teams, including periodic backup, data preprocessing, and machine learning classification.
· Enhanced the overall availability and developer experience, such as implementing automation tests, auto-releasing the newest product version, and migrating the stale framework to the breaking changes version.

· 接受一本有影響力的藝術雜誌採訪，深入探討我對 AI 藝術的概念和想法。
 · 精通透過提示工程和獨特模型生成可控制且風格獨特的圖片。
 · 對整個 t2i 模型進行了微調，訓練了 10 多個 LoRA 模型，包括藝術風格、特定面孔、真實的台灣建築。

· Interviewed by an influential art magazine to delve deeper into my concepts and thoughts for AI Art.
· Proficient in controllable and uniquely styled image generation via prompt engineering and exclusive models.
· Finetuned whole t2i model and trained 10+ LoRA models including art style, specific face, and realistic Taiwan architecture.

Freelance, self-employed

AI 藝術家

• 利用 Modbus 通訊協定整合 10 多個 IoT 裝置。
• 擴寫與重構原本的測試程式並實作新的測試，成功抓出 100 個以上有問題的裝置。
• 理解太陽能產業相關的文件與規格書，例如逆變器、電表、日照計等。
• 利用基礎電學的知識計算與驗證數值是否合理，例如功率、電量與功率因數等。

· Integrated 10+ IoT devices data to the monitoring system via Modbus protocol.
· Refactored original auto pytest scripts and added new testing stages in gitlab CI/CD, detected more than 100 buggy device parsers in the codebase successfully.
· Comprehended the documentation and datasheet of the inverter, power meter, and other common devices in the Photovoltaic industry.
· Calculated and verified values from the device registers by basic electricity concepts, such as power, energy, power factor, etc.

thingnario

後端資料工程實習生

• 領導團隊的資料⼯程師並建立管理制度
• 篩選 330+ 封履歷與⾯試 30+ 位候選⼈

• Set up rules to manage the recruitment of data engineers.
• Reviewed 330+ CVs and interviewed 30+ candidates.

IINUMBERS

部門主管

· 與 4 名資料科學家/資料工程師合作為全台最大的 B2C 電商平台建立銷售預測模型，增加 10%-15% 的準確度。
 · 構思並結合 GCP 雲端服務實現了與雲端服務整合的非同步模型訓練系統，每小時可在固定的 container 環境中產出 9000+ 單日商品模型。
 · 進行了大量的演算法、參數、工具實驗，以提高機器學習模型的準確性和效率，例如 multiprocessing、Spark MLlib 等。 
· 開發動態網路爬蟲，破解滑桿驗證碼等反爬蟲機制，抓取文字、圖片。 
· 設計 schema 並建立高效能的 data pipeline 以處理數億筆的資料，最後再做自動排程讓這些資料以 Power BI 儀表板的形式視覺化並實作一些自訂指標來滿足使用者的需求。

· Collaborated with 4 data scientists/engineers to build the sales forcasting model for the largest B2C ecommerce platform in Taiwan. Improved the accuracy by 10-15% compare to the benchmark. 
· Conceptualized and implemented an asynchronous model training system integrated with cloud service, which can produce 9000+ single-day-commodity models in the fixed container environments per hour.
· Performed numerous algorithms, parameters, tools experiments for improving machine learning model accuracy and efficiency, such as multiprocessing, Spark MLlib, etc.
· Developed dynamic web crawler and cracked the anti-crawler mechanism like slider captcha to scrape text and images.
· Designed database schema and established the high performance data pipeline to handle above 100 million rows of data for automatic Power BI dashboards, also implemented some customized metrics to fulfill users' demands.