Resume

Click here to download the PDF version of my English resume.

Work Experience

• Data Scientist
  Panasonic R&D, Osaka, Japan (Apr 2024 - Present)
  • Contributed to the development of a system by analyzing data and providing insights for product development.
  • Collaborated to develop a system analyzing sleep data, enhancing insights into nighttime awakenings and related physiological/environmental factors.
  • Contributed to the validation of system workflows on Node-RED, ensuring efficient system performance.
• Full-stack Developer and AI Engineer
  Looking Up Co., Ltd., Tokyo, Japan (May 2023 – Mar 2024)
  • Collaborated with a team to develop a web application using Node.js, TypeScript, and MongoDB, which analyzed questionnaire feedback to enhance customer business strategy formulation.
  • Enhanced a text classification model using NLTK, scikit-learn, and pre-trained SpaCy word embeddings to process complex Japanese survey data, achieving 97.02% accuracy and boosting training efficiency.
• Data Engineer (Internship)
  Looking Up Co., Ltd., Tokyo, Japan (Aug 2022 – Mar 2023)
  • Developed a ML model using NLTK, scikit-learn, and SpaCy to filter and analyze Japanese survey data, serving as the foundation model for product development.
  • Developed APIs for ML models with FastAPI and Flask, creating user manuals for team access.

Selected Projects

• LLaVAC: Fine-tuning LLaVA as a Multimodal Sentiment Classifier (Jan 2024)
  • Proposed a method to fine-tune Large Language-and-Vision Assistant (LLaVA) as a classifier for classifying multimodal sentiment labels by designing a prompt to consider unimodal and multimodal labels and generating predicted labels.
  • Outperformed state-of-the-art baselines by up to 7.31% in accuracy and by 8.76% in weighted-F1 in the MVSA-Single dataset.
• Multimodal Sentiment Analysis Using Multiple Labels from Different Modalities (Mar 2023)
  • Collaborated with students to design and implement a sentiment analysis model for social network data, leveraging text, image, and multimodal labels using CLIP, BERT, and RoBERTa. Yielded up to 2% improvement in F1-score over recent models.
  • Attained F1-scores of 74.1% for MVSA-single and 62.0% MVSA-multiple datasets.

Education

• Master of Engineering, Tokyo Institute of Technology, Japan (Apr 2021 - Mar 2023)
  • Major: Information and Communications Engineering
  • Advisor: Prof. Manabu Okumura
  • Thesis: Multimodal Sentiment Analysis Using Multiple Labels from Different Modalities
• Bachelor of Business Administration, South China Agricultural University, Guangzhou, China (Sep 2011 - Jun 2015)
  • Major: Management Information System

Skills

• Technical Skills
  • Programming Languages: Python, C, Java, JavaScript, HTML/CSS, TypeScript
  • ML Toolkits: PyTorch, Hugging Face, OpenCV, Scikit-learn, Spacy, NLTK
  • Tools & Technology: Linux Server, SQL, NoSQL (MongoDB), Docker, Jupyter, GCP, Node-RED
• Languages
  • Chinese: Native
  • English: Advanced, TOEIC: 830/990
  • Japanese: Advanced, JLPT-N1
• Certification
  • IBM Full Stack Software Developer Professional Certificate (Coursera)