Hello! I am Nguyễn Minh Ý, a final-year student majoring in Data Science at Ho Chi Minh City University of Foreign Languages – Information Technology (HUFLIT). I have a strong passion for Deep Learning, Computer Vision, and Natural Language Processing (NLP), and I’m constantly eager to apply these technologies in solving real-world challenges. 👁️🗨️🧠
This repository serves as a collection of my personal projects and AI research, including everything from image recognition to deploying models using Docker. It showcases my hands-on experience in building, testing, and optimizing AI models with various applications.
Here are the primary technologies and tools I work with in my projects:
- Programming Languages: Python, C++, JavaScript, SQL, C#
- Libraries & Frameworks:
- Deep Learning: PyTorch, TensorFlow, Hugging Face Transformers
- Data Processing & Analysis: NumPy, Pandas, Scikit-learn, OpenCV
- Web Development: FastAPI, Flask (for building AI APIs and web services)
- Tools & Platforms:
- Containerization: Docker (for deploying models and creating reproducible environments)
- Development: Jupyter Notebook, VSCode, Git/GitHub
- Cloud Platforms: Google Colab, Kaggle (for collaborative coding and cloud computing)
- Operating Systems: Linux, Windows (familiar with both)
Here are some of my key projects that showcase my skills in AI, deep learning, and model deployment:
- Objective: Developed an AI model to detect clickbait in Vietnamese headlines using large language models (LLMs).
- Techniques Used: Fine-tuning and prompting approaches (Zero-shot, Few-shot, Chain-of-thought).
- Result: Achieved an accuracy of 85.7% using fine-tuning with QLoRA and Llama 3.1 8B model, surpassing zero-shot models in performance.
- GitHub Repo: Clickbait Classification
- Objective: Built a predictive model to forecast coronary heart disease (CHD) using medical data.
- Techniques Used: Logistic Regression, Data Preprocessing, Exploratory Data Analysis (EDA).
- Tech Stack: Python, Scikit-learn, Streamlit (for interactive UI).
- Result: Developed a fully functional web app that predicts the likelihood of heart disease based on user input.
- GitHub Repo: Heart Disease Prediction
- Objective: Implemented PhoBERT for aspect-based sentiment analysis (ABSA) on Vietnamese datasets (hotel and restaurant reviews).
- Techniques Used: Fine-tuning PhoBERT, Multi-task Learning.
- Result: Achieved state-of-the-art results on the VLSP 2018 ABSA dataset, with 99.02% ACD F1-score in the hotel domain.
- GitHub Repo: ABSA VLSP 2018
- Objective: Built a churn prediction system to identify customers likely to cancel their subscription or service.
- Techniques Used: PySpark for big data processing, Random Forest for predictive modeling, Hyperparameter Tuning.
- Tech Stack: Python, PySpark, MongoDB, Scikit-learn, Streamlit (for building interactive dashboards).
- Result: Deployed a scalable churn prediction model capable of handling large datasets.
- GitHub Repo: Customer Churn Prediction
- Objective: Created a system to classify and segment chess pieces from images for a web-based game application.
- Techniques Used: ResNet50 for image classification, Mask R-CNN for object segmentation.
- Tech Stack: Python, PyTorch, OpenCV, Streamlit.
- Result: Developed a real-time web app for chess piece detection, capable of displaying classification results and segmented masks.
- GitHub Repo: Chess Pieces Classification
-
Bachelor of Science in Information Technology (Major: Data Science)
Ho Chi Minh City University of Foreign Languages – Information Technology (HUFLIT)
GPA: 3.5/4.0
English Proficiency: B2 -
4th Place – AI Olympics (OAI HCMC 2025) – Recognized for outstanding performance in an AI competition.
Feel free to reach out or connect with me on any of the following platforms:
- 📧 Email: nguyenminhy7714@gmail.com
- 🔗 LinkedIn: Nguyễn Minh Ý
- 🖥️ GitHub: blanatole
- 📱 Phone: 0989650415