LÊ BẢO MINH DUY

Engineer · AI · Developer

About Me

Welcome to my portfolio. I build practical AI systems and developer tools with a focus on clarity, robustness, and measurable impact.

Skills

Natural Language Processing

Processed Vietnamese & English languages for sentiment analysis, classification, and language generation.

Data Mining

Extracted hidden patterns and rules in complex datasets.

Deep Learning

Implemented neural networks for image processing tasks.

Machine Learning

Developed and optimized predictive models.

Internet of Things

Adept at building end-to-end IoT systems, with expertise in sensor integration.

Web Development

Build simple static website.

Projects

Natural Language Processing

Multimodal Soft Prompt-tuning for Clickbait Detection

Developed a clickbait detection system by combining multiple models—a Graph Attention Network for titles, prompt-tuning for descriptions, and a visual model for images—demonstrating a robust approach to multimodal fusion.

#MSP #Clickbait #RoBERTa-large #CLIP #GAT
View Project

Aspect-Based Sentiment Analysis (ABSA)

A project on fine-grained sentiment analysis for Vietnamese, focusing on Aspect-Based Sentiment Analysis (ABSA) by implementing and comparing the performance of ViBERT and PhoBERT models.

#ABSA #ViBERT #PhoBERT
View Project

Data Mining

Sequential Pattern Mining

Analyzed and implemented various sequential pattern mining algorithms—from frequency-based (GSP, SPADE, PrefixSpan) to advanced utility-based (USpan) and periodicity-based (PFPM, PHM) approaches—to uncover valuable patterns in sequential data.

#SPMF #GSP #SPADE #PrefixSpan #USpan #PHM #PFPM
View Project

Data Analysis Pattern Mining (DAPM)

A project focused on implementing and deploying state-of-the-art algorithms (USpan, HUI-Miner, Two-Phase) for discovering high-utility patterns, with an emphasis on data visualization.

#USpan #HUI-Miner #Two-Phase #Streamlit
View Project

Deep Learning

ASL Recognition

Developed an ASL recognition system by experimenting with various image segmentation and classification algorithms, and deployed the final model into a user-friendly Flask application.

#PyTorch #VGG19 #ResNet50 #DeepLabV3 #U-Net #FCN
View Project

Machine Learning

Fruit Classification

Applied K-NN algorithm, with image feature vectors generated by SIFT, HOG, LBP, and color histograms, and deploying the final model as a web application with Streamlit.

#KNN #SIFT #HOG #LBP #ColorHistograms #GridSearch #Streamlit
View Project

Stock Clustering

Applying K-Means clustering to stock data, optimizing and validating the results with key metrics like Elbow and Silhouette scores.

#Scikit-learn #KMeans #Matplotlib
View Project

Internet of Things

Smart Lighting System

Designed and implemented a smart lighting system with an ESP8266 microcontroller, a photoresistor for ambient light detection, and a 74HC595 IC, all remotely managed through the Blynk platform.

#ESP8266 #IC_74HC595 #Blynk
View Project

Web Development

Booking Concert Ticket

Dynamic web application built with HTML, CSS, and JavaScript, featuring an automated email system (SMTP) and a PHP/MySQL backend, developed locally on XAMPP.

#HTML #CSS #JavaScript #PHP #MySQL #SMTP #XAMPP
View Project

Contact