Ming-Tso Chen 陳明佐
Data Scientist Taipei, Taiwan
- Python, and deep learning frameworks, Torch/Keras
- Extensive experience in natural language processing algorithms and methodologies such as TF-IDF, BERT, etc.
- Developed various programs, Web crawlers, scheduling, reports.
- Proficient in database query operations, PL/SQL
- Experienced in the establishment of data pipelines by ETL tools.
Education
National Chiao Tung University
Master Degree, Master Program of Sound and Music Innovative Technologies in College of Engineering (SMIT)
- Perception Signal Processing Lab
- Thesis:
CNN Based Two-stage Multi-resolution End-to-end Model for Singing Melody Extraction.ICASSP 2019: 1005-1009
Sep. 2016 - Sep. 2018
National Taipei University
Bachelor’s Degree, Electrical Engineering
- Automatic Control
- Senior Project: Smart Speaker Design & Development
Sep. 2012 - Jun. 2016
Work Experience
Data Engineer / Data Scientist ˙ Fubon Life Insurance co. Ltd.
Automation of the selling process voice recording inspection with elderly insure
(高齡投保語音質檢) – Python + Pytorch + Database
- Introduce ASR service, i.e. STT, as the core system
- Process raw texts transformed by STT to align templates.
- Adapt NLP methodologies (TF-IDF, BERT) to make the procedure automatic.
- Shorten the inspector reviewing time up to 30-50%.
Claim operation process optimization ( 理賠流程優化 ) – Python + Pytorch + Database
- Build models with NLP such as TF-IDF, Word Embedding, BERT, ERNIE.
- Recommend/Predict the suitable disease and accident codes.
- Boost up the processing for non-quickly-claim applications.
IFRS 17 adaptation and preparation - Database + SQL + ETL
- Introduce enterprise ETL tool and build up data-flows with ETL tool.
- Re-design data architecture of policy data and integrate it into a data warehouse.
- Implement business logic by SQL script.
- design SQL functions like age calculations.
Other Python related works
- Find out the abnormally log-out records in raw VPN logs.
- Containerized Python scripts, executing automatically according to the schedule.
- Generate Word reports with social network graphs by selenium-based web crawler.
Other experience and jobs
- Data Fabric/Data mesh(數據中台) build-up project (work in progress).
- AI Occupation recommendation (work in progress).
- Speaker verification and speaker diarization Industry-University
co-work with NTU and NCTU. - Enterprise phone system architecture upgrade and transformation.
- Integrate ASR into customer service and customer care.
- Dev-Ops & AutoML platform survey and test.
Dec. 2018 - Present
System Acoustic Intern • Tymphany Acoustic Technology co. Ltd.
- LEAP Simulation
- Simulate the kinds of speakers: SPL/Impendance/Frequency response...etc.
Jul. 2018 - Aug. 2018
AI Algorithm Intern • Information and Communications Research Laboratories
- Music genre transfer
- Using the deep learning architecture algorithm of image processing,
Neural Style Transfer to realize audio style transfer.
- Using the deep learning architecture algorithm of image processing,
Jul. 2017 - Aug. 2017
Skills
- Familiar with Python, PyTorch, Keras, Pandas, Scikit-learn, Librosa, Flask(RESTful API), etc
- Database: Oracle, MSSQL Server, DB2 iSeries(AS400)
- Source Code / Library Management: Bitbucket / JFrog Artifactory
- Application Deployment: Docker
- Experience with natural language processing methodology
(BERT, word2vec, TF-IDF, T5, FM etc.), and deep learning algorithms developments - Other Tools: Trinity (ETL tool)
Qualification
Taiwan AI Academy
Natural Language Processing Technical Program
in Artificial Intelligence.
