楊人豪

工程師

  Taipei, Taiwan

   
 https://github.com/jenhaoyang 

Skills:
Familiar with Python, C++
Use U-Net for segmentation
Use ConNet for suspicious tissue classification
Familiar with Pytorch

Familiar with REST-ful API
Use Flask to make Web API for our inference model
Familiar with Docker, use Container to unify program execution environment
Familiar with MySQL
Use git and GitLab for version control

Programming style:
Follow PEP8 conventions, use pylint for style check
Focus on code Reusable, refactor when code starts to smell bad
Focus on Object-Oriented Programming, use Design Pattern in suitable place
Focus on coding quality and readability


Features:
Get Certification from Deep Learning AI for AI Medical Diagnosis course
TOEIC 855

所用技能:
熟悉Python, C++
使用U-Net進行segmentation
使用ConNet進行結節分類
熟悉Pytorch機器學習框架

熟悉REST-ful API
使用Flask將計算模型包裝成WebAPI讓一般使用者使用
熟悉Docker建立Container, 統一軟體環境
熟悉MySQL
使用git 以及GitLab進行版控

軟體編寫風格:
遵循PEP8軟體寫作風格, 使用pylint 輔助修正
注重程式碼Resuable, 主動尋找程式碼可以加強之處並且重構
注重物件導向設計, 套用設計模式到適合的地方
勇於實驗並找出最佳解, 將實驗失敗視為經驗養份
注重程式碼品質以及可讀性

其他特點:
獲得Deep Learning AI線上課程 AI for Medical Diagnosis認証
多益855分




Experience 工作經歷

Deep Learning for vision Engineer深度學習影像工程師  •  資拓宏宇

May 2020 - Present 五月 2020 - Present

Working in Deep Learning for lung nodules detection in CT scans. Improving Deep Learning pipeline .
Data cleaning, slice CT images, and use Tensorboard to get training metrics.

Use cache function from Python functools to precache data in order to save time from slow IO operation.
Make Web API for Deep Learning model. Use Task Queue to detect massive CT series in background thus avoid blocking the Web API services.

Face Recognition System: Use FaceNet, Pytorch, Flask

負責公司AI醫療影像辨識-肺部CT掃描節結偵測專案開發,優化Deep Learning pipline 尋找肺部可疑節結。從原始CT檔以及醫生標註開始進行資料清洗, 到將照片切割送入模型訓練, 以及使用Tensor board觀察訓練情形。
使用cache預處理資料集,以避免訓練時間被浪費在緩慢的IO操作
同時也為模型製作後端Web API, 並搭配Task Queue 進行大量CT進行偵測,並且在背景工作, 以避免Web API被費時的工作卡住

人臉辨識系統: 以FaceNet, Pytorch, Flask開發人臉辨識系統


Software Engineer 軟體工程師  •  Techman Robot Inc. 達明機器人

November 2018 - May 2020 十月 2018 - 五月 2020

Use C++ to develop 3D robot arm visual environment for motion simulation (known as offline programming). Use MFC and OpenCASCADE.
Use Strategy pattern to design object, make common interface for similar object but different implement for single object.
Use Observer pattern to build message mechanism between objects
Use State machine to build different state for object

After use design pattern for the project, I successfully decouple components in the program. Also, the extensibility also have significant increase.

以C++開發機械手臂3D虛擬環境模擬軟體,搭配MFC以及OpenCASCADE.
為專案導入設計模式. 以Strategy pattern設計物件, 製作物件共同Interface但是各自有不同的Implement.
導入Observer pattern建立物件之間的訊號消息機制
導入State machine, 建立物件不同狀態的不同行為. 

導入設計模式後為專案大幅解耦並且讓專案的擴充性大幅提升

Exchange researcher 交換研究員  •   Fraunhofer IPT in Germany 德國 Fraunhofer IPT

September 2016 - February 2017 九月 2016 - 二月 2017

Calculate wear stone theory wearing rate.

Use Matlab to make plot
進行汽車板金拋光之拋光石模耗理論計算
使用Matlab畫出模擬結果

Education 學歷

2014 - 2017

National Taiwan University of Science and Technology

國立台灣科技大學

Master of Mechanical Engineering

機械工程研究所

Achievements & Honors  證書



AI for Medical Diagnosis Course Certificate

Coursera AI for Medical Diagnosis 修課認証

TOEIC 855




Powered by CakeResumePowered by CakeResume