Experience

CreditQuant AI

Software Engineer (Part-time)

Built a full-stack AI financial analytics MVP for real-time credit insights and document processing, earning Cornell Tech Startup Award funding.

Key Achievements:

  • Developed responsive React and TypeScript dashboard with reusable components for ratio visualization
  • Implemented a secure Node.js backend with AWS Lambda, API Gateway, Cognito, and REST endpoints
  • Integrated RAG pipelines using OpenAI API to normalize financial statements and extract credit metrics
ReactTypeScriptPythonNode.jsAWS LambdaS3CognitoCloudFare

TranscribeGlass

Software Engineering Intern

Developed and maintained an iOS application of a wearable automatic speech recognition device for hearing loss.

Key Achievements:

  • Built iOS app using Objective-C and Swift, achieving 20% increase in performance efficiency
  • Integrated Firebase services, resulting in 35% faster data processing
  • Optimized data pipelines with gRPC and Protobuf, reducing ML model training time by 25%
SwiftObjective-CFirebasegRPCProtobufiOS Development

CITRIS & Banatao Institute

Data Scientist Intern

Led AI research initiative for improving pancreatic cancer detection while managing a team of 10 members.

Key Achievements:

  • Trained hierarchical shrinkage tree algorithms with 15% improved accuracy
  • Built comprehensive dataset of 100,000+ pancreatic cancer patients
  • Created data visualizations using matplotlib, seaborn, and Tableau for stakeholders
Pythonscikit-learnMatplotlibSeabornTableauDeep Learning

Just As Special (JAS)

Data Science Intern

Leveraged AI and NLP technologies to analyze foster care organization data.

Key Achievements:

  • Transcribed 50+ interview podcasts using Google Cloud Speech-to-Text
  • Implemented NLP analysis using TextBlob and NLTK for data structuring
  • Presented data-driven whitepaper at Academic Data Science Alliance conference
Google CloudTextBlobNLTKpandasSpaCyData Analysis

National Academy of Taiwan

Machine Learning Intern

Developed emotion recognition and persona inference models using advanced ML techniques.

Key Achievements:

  • Built models using RoBERTa and XLM-RoBERTa for feature extraction
  • Optimized architectures with GRU, RNN, and LSTMs
  • Enhanced emotion recognition accuracy by 5% using TensorFlow and Keras
PythonTensorFlowPyTorchHugging FaceBERTCOMET