About me

In my current role, I design and implement solutions that streamline inter-department processes, reducing friction in the software development lifecycle. I focus on creating tools that bridge communication gaps and automate repetitive tasks, enabling teams to focus on what they do best.

My technical foundation spans full-stack development, mobile applications, and database systems, with strong proficiency in Python, TypeScript, and C++. I'm driven by the challenge of solving complex problems through elegant, scalable solutions that make development teams more productive.

What I'm doing

  • Web development icon

    Web Development

    Design and build the front-end, back-end, and database of web applications.

    TypeScript Python
    React Node.js
    SQL MongoDB BigTable Redis

  • Machine Learning

    Develop ML applications that involves natural language processing (NLP) and RAG systems capabilities.

    Python Pytorch
    Streamlit Huggingface Langchain Vector Store

  • DevOps

    Build CI/CD pipelines to efficiently automate and streamline software deployment.

    Docker Kubernetes AWS GitLab CI Ansible Jenkins

  • Desktop Application

    Build desktop application that run can locally on Mac and Windows

    Python PyQT

Experience

Work

  1. Platform Engineer - InterSystems

    February 2024 — Present

  2. Software Engineer Intern · Lenovo

    May 2023 — August 2023

  3. Software Engineer Intern · Cathay United Bank

    Oct 2021 - Jan 2022
    • Designed and implemented a proof-of-concept of an automated hybrid cloud machine learning pipeline for banking data residency requirements using AWS SageMaker, CodePipeline, and EventBridge.
    • Created AWS CloudFormation templates that allowed the company to configure the infrastructure resources and deploy the pipeline in less than 10 minutes.
    • Applied GitOps practices with ArgoCD on Kubernetes clusters to automatically sync and monitor applications from git repositories and container registries, increasing efficiency and security for deployment.

  4. Research Assistant · NTOU NLP Lab

    Jan 2021 - Jan 2022
    • Developed text preprocessing pipelines and experimented with Transformer-based models in Taiwanese-Chinese translation, achieving a BLEU score of nearly 90 on translating 10,000+ sentences.
    • Deployed the machine translation models in web application using Streamlit, allowing users to adjust various parameters and observe the resulting translations in real-time.
    • Led a 4-man team to fine-tune and evaluate GPT-2 model capabilities on rephrasing sentences for commercial dialogues.

Education

  1. Rice University

    Master of Computer Science
    Aug 2022 — Dec 2023

    Courses Works:
    - Web Development Design
    - Graduate Design & Analysis of Algorithms
    - Computer System Architecture
    - Database System Implementation
    - Software Engineer Methodology
    - Machine Learning
    - Deep Learning with Vision and Language

  2. National Taiwan Ocean University

    Bachelor of Science in Computer Science
    Sep 2018 — Jan 2022

Projects

  • Owl Go - Social Networking Website

    Web development


    owlgo

    Overview

    The social networking website that offers features cater to users' diverse needs. Functionalities include user registration and authentication, profile management, and the ability for users to follow friends, post and comment on articles, and upload images. Additionally, the website consists of advanced features such as OAuth 2.0 and account linking, which enhance the user experience by providing streamlined login processes by connecting existing accounts with third-party accounts.


    Tech Stack

    TypeScript React Node.js Express.js MongoDB Jest



  • Taiwanese-Chinese Machine Translation

    Machine Learning


    machine translation

    Overview

    This app generates Machine Translation from Taiwanese (Hokkien) to Traditional Chinese characters using the Transformer architecture. We also implemented Beam Search and Length penalty to improve performance.


    Tech Stack

    Python Pytorch Streamlit



  • CLIPGPT-ImageCaptioner

    Machine Learning


    image captioner

    Overview

    This app utilizes OpenAI's GPT-2 and CLIP models to generate image captions. The model architecture was inspired by ClipCap: CLIP Prefix for Image Captioning, which uses CLIP encoding as prefix and fine-tune GPT-2 model to generate the caption.


    Tech Stack

    Python Pytorch Streamlit Huggingface



  • Chinese Language Sample Analysis System for Children

    Desktop Application


    CLSA

    Motivation

    The system was developed based on the book "Chinese Language Sample Analysis". Our goal is to provide speech therapists with a more efficient system for recording and analyzing children's speech samples. However, in the past, therapists had to transcribe the content of the audio files verbatim into an online word segmentation system and use additional software to analyze the results and record them in a book, making language sample analysis time-consuming and laborious.


    Implementation

    The system was developed using a Python GUI (PyQt5). We leveraged CKIP Transformers to perform word segmentation and POS tagging, combined with the Microsoft Speech-to-Text SDK for speech recognition. We also use MongoDB to help therapists with data management. During development, we held multiple discussions with the therapists to design for the clinician's focus, including the calculation of "mean length of utterances", and "vocabulary diversity".


    Tech Stack

    Python PyQT MongoDB Azure



Blog

Contact

Contact Form