CakeResume 找人才
進階搜尋
On
4 到 6 年
6 到 10 年
10 到 15 年
15 年以上
Tainan City, Taiwan
Technical Project Manager/技術專案管理 @CM Visual Technology Corporation/微采視像科技股份有限公司2016 ~ 現在
・
AI工程師、機器學習工程師、深度學習工程師、資料科學家、Machine Learning Engineer、Deep Learning Engineer、Data Scientist
超過一年
data science. Accustomed to communicating and failure analyzing with cross department. Tainan City, Taiwan Skill Experience 5yrs Product development and validation 3yrs FAE to client 3yrs Display development Optic Display optics and design Wide-view technology Polarizer and Thin film process Knowledge of Color theory, Optic, Crystal optic Data science Statistic Python, R, SQL DFSS Knowledge of ML/DL, Data modeling and visualization Personality Traits Motivated learner Accustomed to communicating Flexible and open-mind Language Chinese - Native English - Intermediate Work Experience Technical Project Manager CM Visual Technology Corporation •Present Responsible for: 1. New product development project
就職中
全職 / 對遠端工作有興趣
6 到 10 年
National Sun Yat-Sen University/國立中山大學
・
Optoelectronic Engineering/光電工程
就職中
・
目前會考慮了解新的機會
全職 / 對遠端工作有興趣
6 到 10 年
HHL LEIPZIG GRADUATE SCHOOL OF MANAGEMENT – Top 100 Global MBA by The Economist
・
Management, Entrepreneurship, Strategy
資深經理 @緯創資通2021 ~ 現在
・
Technical Manager
半年內
National Taiwan University of Science and Technology - Master of Information Engineering Experience: 2010 Wistron - Software Engineer 2015 Wistron - ML/DL Image Processing Engineer 2019 Wistron - Technical Manager 2020 MIT - Computer Science and Artificial Intelligence Laboratory (CSAIL) Visiting Engineer Competition: 2016 Ministry of Economic Affairs Bureau of Industry – Mastering the Data Context Hackathon Competition, Data Marketing Award 2017 Kaggle: The Nature Conservancy Fisheries Monitoring – bronze medal <6% 2018 Wistron Capital Entrepreneurship Competition 3rd place - Baby Guardian In 2020, I was fortunate to be a Visiting Engineer at Massachusetts Institute of Technology-Computer Science & Artificial Intelligence Laboratory for six months, to
就職中
全職 / 對遠端工作有興趣
10 到 15 年
National Taiwan University of Science and Technology
・
Master's degree Computer Science and Information Engineering
半導體研發工程師 @ UMC2015 ~ 現在
・
三個月內
20% to almost zero by optimizing plasma uniformity in ULK deposition. Troubleshooting in dielectric process for 14 nm and 22 nm mass product. Optimized ULK film for 28 nm shrinkage to 22 nm technology. Developed high compressive TEOS for wafer bonding. ( < -350 MPa) Developed Ar carrier gas in TEOS deposition for helium reduction. Education History National Tsing Hua University (NTHU), Taiwan,Ｍaster of Science in Chemistry National Kaohsiung Normal University, Taiwan,Bachelor of Science in Chemistry Skill Specializing in ULK/TEOS/ALD processes. Experience with LAM and AMAT t...
全職 / 對遠端工作有興趣
4 到 6 年
清華大學
・
化學就職中
・
目前沒有興趣尋找新的機會
全職 / 對遠端工作有興趣
6 到 10 年
國立清華大學
・
社會學, 哲學
缺陷分析工程師 @TSMC2014 ~ 2017
・
資料科學家/機器學習工程師
超過一年
problems, familiar with the way that defect is discovered and the mechanism of defect formation. Also experienced on develop defect analysis system with IT and machine learning engineer. For job skills/tools, learned to program by VBA, SQL and some python because of the large amount of data analysis at work, and also try to apply machine learning to work. Currently working and studying hard to become a data scientist. 資料科學家/機器學習工程師 Taipei Special Municipality，TW [email protected] 工作經
全職 / 對遠端工作有興趣
4 到 6 年
清華大學
・
物理系最輕量、快速的招募方案，數百家企業的選擇
搜尋履歷，主動聯繫求職者，提升招募效率。
- 瀏覽所有搜尋結果
- 每日可無限次數開啟陌生對話
- 搜尋僅開放付費企業檢視的履歷
- 檢視使用者信箱 & 電話
搜尋技巧
1
嘗試搜尋最精準的關鍵字組合
資深 後端 php laravel
2
將須完全符合的字詞放在雙引號中
"社群行銷"
3
在不想搜尋到的字詞前面加上減號，如果想濾掉中文字，需搭配雙引號使用 (-"人資")
UI designer -UX
職場能力評價定義
專業技能
該領域中具備哪些專業能力（例如熟悉 SEO 操作，且會使用相關工具）。
問題解決能力
能洞察、分析問題，並擬定方案有效解決問題。
變通能力
遇到突發事件能冷靜應對，並隨時調整專案、客戶、技術的相對優先序。
溝通能力
有效傳達個人想法，且願意傾聽他人意見並給予反饋。
時間管理能力
了解工作項目的優先順序，有效運用時間，準時完成工作內容。
團隊合作能力
具有向心力與團隊責任感，願意傾聽他人意見並主動溝通協調。
領導力
專注於團隊發展，有效引領團隊採取行動，達成共同目標。
專業背景
目前狀態
就職中
求職階段
目前沒有興趣尋找新的機會
專業
機器學習工程師, 數據科學家
產業
人工智慧 / 機器學習, 大數據, 軟體
工作年資
2 到 4 年
管理經歷
我有管理 1～5 人的經驗
技能
語言能力
English
・
專業
求職偏好
希望獲得的職位
Machine Learning Engineer
預期工作模式
全職
期望的工作地點
台灣台北, 台灣台南市, 美國加利福尼亞洛杉磯, 德國慕尼黑, 加拿大安大略多倫多
遠端工作意願
對遠端工作有興趣
接案服務
否
學歷
學校
NCKU, Master Degree
主修科系
Computer Science
Kai-Chou, Yang
As a Kaggle Competition Master and a winner of international data science challenges, I am experienced in machine learning, deep learning and related frameworks such as PyTorch.
My research focuses on natural language processing (NLP), where I have released 11 open-source projects such as MianBot (700+★ on Github) and presented certain academic papers on top conferences like ACL, AAAI, CIKM, and WSDM.
International Awards
For the following achievements, I am the first author as well as the team leader.
2nd Place, CIKM Cup: Cross-lingual Short-text Matching Challenge
- Proposed two densely-connected architectures, CPRNN and DACNN, for sentence pair modeling.
- Fused semantic features from different levels to create diversity intra-models.
- The solution has been oral presented on CIKM 2018 in Turin, Italy.
3rd Place, WSDM Cup: Fake News Classification Challenge
- Implemented various NLI networks like ESIM and injected world knowledge using BERT.
- Proposed a disagreement-aware model based on the single-word attention.
- The paper has been oral presented on WSDM 2019 in Melbourne, Australia.
4th place, Google AI: Gendered Pronoun Resolution Competition
- Leveraged the information redundancy from BERT and extracted features from the optimal layer.
- Proposed a multi-heads Siamese semantic scorer for answer selection.
- The paper has been presented on ACL 2019 in Florence, Italy.
Kaggle Competition Master, Ranked top 0.2% (233/114,366)
- Top 1% (4/838), Gendered Pronoun Resolution Competition.
- Top 1% (27/4,550), Toxic Comment Classification Challenge.
- Top 3% (30/1,449), CareerCon 2019 - Help Navigate Robots.
- Top 4% (103/3,165), Jigsaw Unintended Bias in Toxicity Classification.
- Top 6% (223/3,633), CommonLit Readability Prize.
- Top 10% (384/3,946), TalkingData AdTracking Fraud Detection Challenge.
Work Experience
Taiwan AI Labs, Machine Learning Engineer, Sep 2019 ~ Now
Question Answering System
- Propose a conditional question generator with mT5 for controllable QA data augmentation and as the base of dense retrieval, which improves recall@50 from baseline model by 16%.
- Build a generative pseudo labeling pipeline using a open-domain passage retriever and machine reader, which improve the nDCG@10 by 4.2 - 9.7, on various domains.
- Build an efficient passage re-ranker based on tiny-bert with a time-series based clustering framework for effective negative passage sampling.
- Leverage FinBERT on QA analysis and slot filling for fintech dialogue system.
Natural Language Understanding
- Implement a document encoder with self-contrastive learning and a document clustering algorithm, which is scalable for million scale of streaming data.
- Implement a GROVER-like generator as the backbone for topic detection, article rewriting, and tag generation.
- Propose a semi-automatic framework for fake-news identification, which gathers evidence from event properties, user behavior and textual features.
- Propose a SOTA Chinese typo correction system based on a boosting loop of automatic speech recognition and text to speech for weak supervision.
- Build a general-purpose NLP training pipeline for team use involving data augmentation, data regularization, and unsupervised domain adaptation.
Education
Master in Department of Computer Science, NCKU GPA: 4.30
- Honorary member of the Phi Tau Phi Scholastic Honor Society. (Ranked 1st among all graduates.)
- As a teaching assistant for Introduction to Data Science, Data Mining and Discrete Mathematics.
- As a speaker / teaching assistant for introduction lectures of machine learning.
Bachelor in Department of Computer Science, NCKU GPA: 3.92
- Academic excellence awards 2016.
- Academic excellence awards 2015.
- Honorable mention on the graduation exhibition.
- Research assistant on a question answering system project for the Ministry of Science and Technology.
Side Projects
I list some of my project experiences. You can refer to my Github for the other interesting ideas.
Mianbot
- Got 700+ stars and 200+ forks on Github.
- Implemented the hierarchical keywords matching using word2vec.
- Implemented the IR-based searching module to support chit-chat.
- Allow user to define customized scenarios with JSON.
- The extracted QA pairs were released in PTT-Gossiping-Dataset, a widely-used Chinese chit-chat corpus.
NCKU Smart-Life LineBot
- A Linebot that helps solve trivial matters such as restaurant recommendation.
- The dialogue system is based on LUIS for intent classification.
- The backend was built with Django / Flask (new version) and host on Heroku.
- The backend is connected with Line server using the web API.
Knowledge & Skills
- General Machine Learning
- Classification, Regression, Clustering, Boosting, Feature Engineering.
- Natural Language Processing
- Sentence Pair Modeling: Natural language Inference, Machine Reading Comprehension, Sentence Similarity
- Text Classification / Regression / Clustering
- Deep contextual representation (ELMO / BERT / XLNet / ELECTRA / RoBERTa / ERINE2.0 / BigBird / T5)
- Recommendation System
- Factorization: Matrix Factorization, Factorization Machine, DeepFM
- Graph Embedding: DeepWalk, Node2Vec, item2Vec
Publication
- Fake News Detection as Natural Language Inference. Kai-Chou Yang; Timothy Niven; Hung-Yu Kao. WSDM Cup 2019
- Fill the GAP: Exploiting BERT for Pronoun Resolution. Kai-Chou Yang; Timothy Niven; Tzu Hsuan Chou; Hung-Yu Kao. ACLWS'19
- Generalize Sentence Representation with Self-Inference. Kai-Chou Yang; Hung-Yu Kao. AAAI 2020
- The Prevalence and Impact of Fake News on COVID-19 Vaccination in Taiwan: A Retrospective Study of Digital Media. Yen-Pin Chen; Yi-Ying Chen; Kai-Chou Yang; Feipei Lai; Chien-Hua Huang; Yun-Nung Chen; Yi-Chin Tu. JMIR
Kai-Chou, Yang
As a Kaggle Competition Master and a winner of international data science challenges, I am experienced in machine learning, deep learning and related frameworks such as PyTorch.
My research focuses on natural language processing (NLP), where I have released 11 open-source projects such as MianBot (700+★ on Github) and presented certain academic papers on top conferences like ACL, AAAI, CIKM, and WSDM.
International Awards
For the following achievements, I am the first author as well as the team leader.
2nd Place, CIKM Cup: Cross-lingual Short-text Matching Challenge
- Proposed two densely-connected architectures, CPRNN and DACNN, for sentence pair modeling.
- Fused semantic features from different levels to create diversity intra-models.
- The solution has been oral presented on CIKM 2018 in Turin, Italy.
3rd Place, WSDM Cup: Fake News Classification Challenge
- Implemented various NLI networks like ESIM and injected world knowledge using BERT.
- Proposed a disagreement-aware model based on the single-word attention.
- The paper has been oral presented on WSDM 2019 in Melbourne, Australia.
4th place, Google AI: Gendered Pronoun Resolution Competition
- Leveraged the information redundancy from BERT and extracted features from the optimal layer.
- Proposed a multi-heads Siamese semantic scorer for answer selection.
- The paper has been presented on ACL 2019 in Florence, Italy.
Kaggle Competition Master, Ranked top 0.2% (233/114,366)
- Top 1% (4/838), Gendered Pronoun Resolution Competition.
- Top 1% (27/4,550), Toxic Comment Classification Challenge.
- Top 3% (30/1,449), CareerCon 2019 - Help Navigate Robots.
- Top 4% (103/3,165), Jigsaw Unintended Bias in Toxicity Classification.
- Top 6% (223/3,633), CommonLit Readability Prize.
- Top 10% (384/3,946), TalkingData AdTracking Fraud Detection Challenge.
Work Experience
Taiwan AI Labs, Machine Learning Engineer, Sep 2019 ~ Now
Question Answering System
- Propose a conditional question generator with mT5 for controllable QA data augmentation and as the base of dense retrieval, which improves recall@50 from baseline model by 16%.
- Build a generative pseudo labeling pipeline using a open-domain passage retriever and machine reader, which improve the nDCG@10 by 4.2 - 9.7, on various domains.
- Build an efficient passage re-ranker based on tiny-bert with a time-series based clustering framework for effective negative passage sampling.
- Leverage FinBERT on QA analysis and slot filling for fintech dialogue system.
Natural Language Understanding
- Implement a document encoder with self-contrastive learning and a document clustering algorithm, which is scalable for million scale of streaming data.
- Implement a GROVER-like generator as the backbone for topic detection, article rewriting, and tag generation.
- Propose a semi-automatic framework for fake-news identification, which gathers evidence from event properties, user behavior and textual features.
- Propose a SOTA Chinese typo correction system based on a boosting loop of automatic speech recognition and text to speech for weak supervision.
- Build a general-purpose NLP training pipeline for team use involving data augmentation, data regularization, and unsupervised domain adaptation.
Education
Master in Department of Computer Science, NCKU GPA: 4.30
- Honorary member of the Phi Tau Phi Scholastic Honor Society. (Ranked 1st among all graduates.)
- As a teaching assistant for Introduction to Data Science, Data Mining and Discrete Mathematics.
- As a speaker / teaching assistant for introduction lectures of machine learning.
Bachelor in Department of Computer Science, NCKU GPA: 3.92
- Academic excellence awards 2016.
- Academic excellence awards 2015.
- Honorable mention on the graduation exhibition.
- Research assistant on a question answering system project for the Ministry of Science and Technology.
Side Projects
I list some of my project experiences. You can refer to my Github for the other interesting ideas.
Mianbot
- Got 700+ stars and 200+ forks on Github.
- Implemented the hierarchical keywords matching using word2vec.
- Implemented the IR-based searching module to support chit-chat.
- Allow user to define customized scenarios with JSON.
- The extracted QA pairs were released in PTT-Gossiping-Dataset, a widely-used Chinese chit-chat corpus.
NCKU Smart-Life LineBot
- A Linebot that helps solve trivial matters such as restaurant recommendation.
- The dialogue system is based on LUIS for intent classification.
- The backend was built with Django / Flask (new version) and host on Heroku.
- The backend is connected with Line server using the web API.
Knowledge & Skills
- General Machine Learning
- Classification, Regression, Clustering, Boosting, Feature Engineering.
- Natural Language Processing
- Sentence Pair Modeling: Natural language Inference, Machine Reading Comprehension, Sentence Similarity
- Text Classification / Regression / Clustering
- Deep contextual representation (ELMO / BERT / XLNet / ELECTRA / RoBERTa / ERINE2.0 / BigBird / T5)
- Recommendation System
- Factorization: Matrix Factorization, Factorization Machine, DeepFM
- Graph Embedding: DeepWalk, Node2Vec, item2Vec
Publication
- Fake News Detection as Natural Language Inference. Kai-Chou Yang; Timothy Niven; Hung-Yu Kao. WSDM Cup 2019
- Fill the GAP: Exploiting BERT for Pronoun Resolution. Kai-Chou Yang; Timothy Niven; Tzu Hsuan Chou; Hung-Yu Kao. ACLWS'19
- Generalize Sentence Representation with Self-Inference. Kai-Chou Yang; Hung-Yu Kao. AAAI 2020
- The Prevalence and Impact of Fake News on COVID-19 Vaccination in Taiwan: A Retrospective Study of Digital Media. Yen-Pin Chen; Yi-Ying Chen; Kai-Chou Yang; Feipei Lai; Chien-Hua Huang; Yun-Nung Chen; Yi-Chin Tu. JMIR