Phan Duc Nha

Phan Duc Nha

Computer Science Student

Researching state-of-the-art AI methods in Vision-Language Models, Reinforcement Learning, and Deep Representation Learning at HCMC University of Technology

About Me

As a Computer Science student at Ho Chi Minh City University of Technology (HCMUT), I developed strong programming proficiency in Python, Java, and C++, alongside a solid foundation in Machine Learning, Deep Learning, and Multimodal Analysis. My academic coursework and independent research cultivated my passion for state-of-the-art AI methods, particularly in Vision-Language Models and Reinforcement Learning.

I'm a young person with a deep passion for exploring and experimenting with artificial intelligence. I've built several research projects focusing on NLP and Computer Vision. I'm always eager to learn new things, improve my skills, and work with others to solve challenging problems. My long-term goal is to become an AI engineer and help create technologies that make a real impact.

Languages

Vietnamese (Native)

English (Fluent)

Technical Skills

Python Java C++ JavaScript SQL

Research Interests

Vision-Language Models

Multimodal learning and alignment between visual and textual data

Reinforcement Learning & Alignment

Preference optimization and policy learning for AI systems

Deep Representation Learning

Learning meaningful embeddings from complex data

Signal Processing & Audio AI

Deep learning on audio signals and speech processing

Featured Projects

2024

ChartQA: Vision-Language Alignment

Parameter-Efficient Fine-Tuning with QLoRA for Chart Question Answering. Fine-tuned Projection layers and Decoder for Chain-of-Thought reasoning.

Qwen3-VL SFT GRPO Vision-Language
View on GitHub
In Process

RL-based Sequential Image Quality Diagnosis

We study image quality assessment as a sequential decision-making problem. The agent incrementally inspects image regions to make a quality diagnosis, balancing perceptual accuracy and inspection cost using preference-based reinforcement learning.

Reinforcement Learning Image Quality Assessment Sequential Decision Making Preference-based RL
View on GitHub
In Process

Speech Extraction: Deep DSP

CNN/RNN-based speech extraction using SI-SDR loss. Speaker Encoder for voice embeddings and Extraction Network for signal separation.

CNN RNN SI-SDR DSP
View on GitHub

Education

HCMC University of Technology

2023 - Present

BSc in Computer Science

Pursuing Bachelor's degree with focus on Artificial Intelligence and Machine Learning

Get In Touch

I'm always open to discussing research opportunities, collaborations, or just chatting about AI!

Location

Ho Chi Minh, Vietnam