I'm an AI researcher passionate about developing intelligent systems that push the boundaries of computer vision and deep learning. Currently pursuing my Master's degree in Computer Science at Shibaura Institute of Technology in Tokyo, supported by the prestigious MEXT Government Scholarship.
Research Focus
Computer vision, deep learning, and practical AI systems for real-world impact.
Current Goal
Bridging cutting-edge research and deployable intelligence for robust applications.
Education
Master's Degree in Computer Science
MEXT ScholarShibaura Institute of Technology
Tokyo, Japan
- Government MEXT full-ride scholarship recipient
Engineer Degree in Information Technology
Hanoi University of Science and Technology
Hanoi, Vietnam
- GPA: 3.51 / 4.0 | Batch ranking: Top 10%
- Thesis: Person Monitoring on Edge Devices using Computer Vision (9.5/10)
- Relevant courses: Probability Theory (A), Machine Learning and Data Mining (A), Computer Vision (A)
High School Diploma
High School for Gifted Students - Hanoi National University of Education
Hanoi, Vietnam
- Specialization: Mathematics
- GPA: 9.3/10
Experience
AI Researcher · GSV Dtech Technology Joint Stock Company
Hanoi, Vietnam
- Designed and implemented LLM-based Text-to-SQL pipelines, translating natural language queries into executable database queries, with focus on accuracy and robustness.
- Fine-tuned embedding models for customer-service chatbots, achieving ~50% improvement in top-5 retrieval accuracy; experimented with prompt design and embedding optimization.
AI Researcher · DAYONE Joint Stock Company
Hanoi, Vietnam
- Implemented key information extraction model on invoices based on graph neural networks for more than 30 types of invoices.
- Developed fraud detection models to identify screen-captured images.
- Collaborated with the team to deploy trained models on production.
Research Intern · VNG Corporation
Hanoi, Vietnam
- Generated synthesis data for document corner regression problems using basic image processing techniques.
- Trained OCR model to recognize characters on identity cards and achieved 99.9% word accuracy.
- Performed feature extraction of words on documents based on their shapes and trained a transformer-based model to classify splicing words.
- Trained a classification model with EfficientNetB0 architecture to normalize rotated documents and achieved 99.9% accuracy.
- Used deep metric learning to train a deep learning model for the signature verification task.
Research Laboratory Member · BK.AI - International Research Center for AI
Hanoi University of Science & Technology
- Created the Static Hand PosturE (SHAPE) dataset with more than 34,000 images of 32 different hand gestures from 20 people.
- Created the Dynamic hAnd gesTurE (DATE) dataset with 13,500 videos of 27 different hand gestures from 22 people.
- Customized MobileNetV2 model to create a more lightweight architecture and achieved 90% accuracy on the SHAPE dataset.
- Designed a lightweight model to classify videos of hand gestures based on 3D convolutional neural networks and MobileNetV2 architecture.
Publications
A lightweight architecture for hand gesture recognition
Dang, T. L., Pham, T. H., Dang, Q. M., & Monet, N.
Multimedia Tools and Applications, 82(18), 28569-28587
Person re-identification on lightweight devices: end-to-end approach
Dang, T. L., Pham, T. H., Le, D. L., Tran, X. T., Le, H. N., Nguyen, K. H., & Trinh, T. T. N.
Multimedia Tools and Applications, 83(29), 73569-73582
DATE: a video dataset and benchmark for dynamic hand gesture recognition
Dang, T. L., Pham, T. H., Dao, D. M., Nguyen, H. V., Dang, Q. M., Nguyen, B. T., & Monet, N.
Neural Computing and Applications, 36(28), 17311-17325
Honours & Awards
Gold Award & Most Efficient AI Algorithm Award
Global AI Challenge for Building E&M Facilities 2025
Led a team that achieved 1st place on the leaderboard based on official evaluation metrics. Our proposed AI solution was recognized for both overall excellence and algorithmic efficiency, earning a total prize of USD 10,000 in cash.
The challenge focused on developing AI models to predict the cooling demand of commercial buildings, a critical task for improving building energy efficiency and sustainable facility management.
Grand Prize
Global AI Challenge for Building E&M Facilities 2022
Awarded one of five Grand Prizes, receiving a USD 13,000 cash prize.
The competition featured 126 teams from more than 10 regions worldwide, with over USD 200,000 in total prizes. Hosted by EMSD, HKSAR Government and Guangdong Provincial Association for Science and Technology.
Top 4
KALAPA Bytebattles 2023 Challenge
Ranked in the top 4 of the Vietnamese Hand-written Text Recognition track.
KALAPA Bytebattles 2023 is a highly competitive AI challenge in Vietnam that attracted participation from over 500 teams across the country.
First Prize
University Student Research Conference 2023
Awarded the first prize at the University Student Research Conference 2023, Hanoi University of Science and Technology.
Consolation Prize
The Ministerial Science and Technology Award for Students 2023
Awarded the consolation prize of the Ministerial Science and Technology Award for Students in Higher Education Institutions in 2023.