RESEARCH

My engagement with research stems from a desire to move beyond theory and ask how advanced technologies can respond to real human needs. I am particularly drawn to projects where artificial intelligence, creativity, and accessibility intersect.

A human performing sign language gesture

A Deep Learning-Based Multimodal Communication Framework for Assisting Deaf–Mute Individuals via Computer Vision

Abstract: Deaf–mute individuals face significant communication barriers due to the gap between sign language and spoken language. This study proposes a deep learning-based multimodal communication framework designed to bridge this gap for Vietnamese Sign Language (VSL). The system integrates computer vision and natural language processing techniques, employing CNN–LSTM and 3D-CNN models to recognize hand gestures from video sequences. The recognized signs are then converted into Vietnamese text and speech using a Text-to-Speech (TTS) module. A user-friendly software interface was developed to simulate two-way communication between deaf– mute individuals and normal speakers in real time. Experimental results demonstrated an accuracy of over 85% in sign recognition with low latency, confirming the feasibility and potential application of the proposed framework in real-world scenarios.
Keywords: Deep Learning; Computer Vision; CNN–LSTM; Text-to-Speech (TTS); Vietnamese Sign Language (VSL)

Read full paper

Soundverse - an AI music generation platform

Summary: This is an AI music generation platform I co-founded as an AI engineer. What began as an exploration of generative models evolved into a full-stack system capable of transforming textual prompts into multi-genre music. By integrating synthetic vocals, automatic lyric generation, and style-learning mechanisms that adapt to user-provided audio samples, Soundverse was designed to make music creation accessible to non-experts. This project deepened my understanding of human–AI interaction and reinforced my interest in designing systems that empower creativity rather than replace it.

View Fanpage

View website

Summary: I designed and developed an autonomous warehouse robot using ESP8266, capable of recognizing packages through QR and barcode scanning and navigating storage workflows independently. The project involved end-to-end development-from electronic circuit design and sensor integration to mechanical modeling, 3D printing, and physical assembly.
To enable real-time monitoring and control, I implemented an API-based IoT web server, allowing data visualization and system interaction remotely. Building LogiBot strengthened my appreciation for robustness, testing, and iteration in engineering, and reinforced my interest in developing intelligent systems that operate reliably beyond laboratory settings.

LogiBot - Autonomous Warehouse Robot

View Youtube