Shih-Wen Tsou

- With more than 5 years of experience in Data Analysis, Machine Learning and Deep Learning, familiar with Modeling, Data Analysis, Image Processing, Machine Learning, and Deep Learning.

  Taipei City, Taiwan      

WORK EXPERIENCE

Lead Data Scientist / Full Stack Data Scientist, Vinnovation Network, Taipei, Taiwan

Data Engineering / Data Analysis

  • Spearheaded the development of a fully automated data integration pipeline that aggregated diverse data sets into a S3 Data Lake.
  • Successfully integrated a range of data sources, including real-time data feeds from AWS Redshift and DocumentDB, as well as batch processes to import traditional CSV files.
  • Utilized Databricks for large-scale data processing, leveraging its Spark capabilities to efficiently transform and aggregate incoming data streams.
  • With the combined power of Databricks and AWS Lambda, ensured unparalleled data consistency, quality, and preparedness for sophisticated analytics and reporting.
  • Utilized Databricks and Airflow to run extensive data profiling tasks, analyzing data patterns and identifying potential quality issues before they reached the Databricks Delta Lake.
  • Established robust guardrails using the combined might of AWS Lambda, Apache Airflow and Databricks, ensuring that data stored in the DataBricks Delta Lake consistently met the highest quality benchmarks.

MLOps / Machine Learning / Data Science

  • Utilized Databricks to build a LightGCN-based recommendation system, fine-tuning for precise content delivery. Monitored model versions with MLflow, ensuring continuous integration.
  • Seamlessly merged our recommendation system with Databricks Delta Lake, maintaining a high-quality data influx and elevating system performance.
  • Developed a comprehensive MLOps service on Databricks, spanning from preprocessing to deployment. Leveraged automation tools to swiftly adapt models to new data.
  • Developed an API service used as an internal tool within the company, leveraging OpenAI Whisper for automatic speech recognition and ChatGPT for intelligent language translation, resulting in approximately an 80% reduction in manpower and time costs.
  • Fine-tuned a Llama-based Large Language Model (LLM) and, by integrating Langchain and Pinecone, optimized the search experience on our website.

Apr. 2022 - Jul. 2023

Machine Learning Researcher, Lab. for Cloud Dynamics and Modeling, National Taiwan University, Taipei, Taiwan    

  • Developed a model using U-net to detect ship tracks in satellite images, resulting in an 80% times savings.

  • Cooperated with multiple domain experts, including Atmospheric Science and Environmental Science, to solve problems with machine learning techniques.

  • Developed a model to classify Typhoon tracks with 96.5% accuracy rate, where the traditional method is about 80%.

  • Configured and managed a GPU-enforced workstation for the lab members to execute High Performance Computing (HPC) tasks.

Sep. 2021 - Mar. 2022

Data Scientist, Vizuro, Taipei, Taiwan

  • Developed an end-to-end pipeline to detect Breast Cancer in 3D Breast MRI images, encompassing data storage, data pre-processing, and detection model building.

  • Leveraged the power of deep learning algorithms, fusing them with insights gathered from medical research, to refine and augment the performance of our breast cancer diagnosis model.

  • Deployed the Deep-learning Breast Cancer Detection model integrated to the hospital PACS system.

  • Developed a model to segment 3D breast MR images and deployed it to ImageJ to expedite annotation and reduce labeling time.

May. 2019 - Sep. 2021

EDUCATION

Sep. 2016 - Jan. 2019

National Taiwan University

Master of Science in Atmospheric Sciences

Sep. 2011 - Jun. 2016

National Taiwan University

Bachelor of Science in Atmospheric Sciences

SKILL

Programming: Python, R, C/C++, GO, Matlab, Fortran

Machine Learning:

  • Traditional: Random Forest, XGBoost, K-means, DBSCAN, PCA, t-SNE
  • Deep Learning: CNNs, RNNs, transformers, GPT, Fast R-CNN series, U-Net, DCGANs, Whisper, Explainable AI techniques

Databases: PostgreSQL, MySQL, MongoDB

Data Engineering: Databricks, PySpark, Airflow, Airbyte

Cloud (AWS): Lambda, S3, EC2, Personalize, VPC

Side Projects


  • Lightning Prediction with Deep Learning and explain the model with physical methods.

  • Learning to generate the Manhattan building with Deep Convolutional GAN from OpenStreeMap building model.

  • Predicting short-term stock market price trends with Machine Learning.

  • Build an Investment Portfolio machine with a Rebalancing Strategy from scratch.