About
About Me
I’m Tran Gia Huy, a 3rd-year Data Science student at the University of Science, VNU-HCM (GPA: 3.68/4, Ranked Top 3 in Semester I 2024–25), and an emerging AI researcher pushing the boundaries of natural language processing,computer vision and machine learning.
My passion lies in crafting lightweight, mathematically rigorous AI solutions that bridge theory and real-world impact. My work on event-based eye tracking, published at CVPRW 2025, achieved a groundbreaking Euclidean error of 1.61 on the 3ET+ benchmark, leveraging innovative spatiotemporal modeling and custom data augmentation.
Key milestones in my journey:
- Academic Encouragement Scholarship: Awarded to Top 4/105 Data Science students at VNU-HCM for academic excellence in Semester I 2024–25.
- CVPRW 2025 Publication: Co-authored a groundbreaking paper on event-based eye tracking, achieving a 1.61 Euclidean error on the 3ET+ benchmark.
- COMAP MCM 2025 Successful Participant: Led Olympic medal predictions with Prophet forecasting and causal inference, representing University of Science, VNU-HCM on an international stage.
- LLM-Powered Real Estate System: Engineered a Vietnamese NLP pipeline with Gemini API, boosting XGBoost accuracy by 18% for property valuation.
- Open-Source Impact: Contributed PyTorch-based vision pipelines, empowering the global AI community.
Fluent in Python, PyTorch, and ML deployment, I thrive on transforming complex challenges into scalable, intelligent systems. Let’s collaborate to shape the future of AI innovation!
Experience
My journey in academiaBachelor of Data Science
Major: AI Engineer | University of Science, VNU-HCMAcademic Encouragement Scholarship
Top 4/105 Data Science StudentsResearch Paper Publication
CVPRW 2025 (Event-Based Vision Workshop)COMAP Mathematical Contest in Modeling
Problem C: Data Insights | VirtualCertificate of Merit
Khanh Hoa Provincial People’s CommitteeSkills
My technical & other skillsData Science & AI
3+ Years XPMachine Learning
90%Deep Learning
90%Computer Vision
90%Natural Language Processing
90%Data Analysis
85%Programming Languages
3+ Years XPPython
95%MATLAB
80%C++
75%Libraries & Frameworks
3+ Years XPPyTorch
90%Pandas
85%NumPy
85%Scikit-learn
80%TensorFlow
75%OpenCV
70%BeautifulSoup
70%Tools
3+ Years XPGit
90%LaTeX
85%Jupyter Notebook
90%Visual Studio Code
85%Kaggle
80%LLM API
70%Soft Skills
3+ Years XPTeamwork
90%Logical Thinking
95%Leadership
85%Critical Thinking
90%Problem Solving
95%Scientific Writing
85%Research
My research publications, conference contributions, & patents
Dual-Path Enhancements in Event-Based Eye Tracking: Augmented Robustness and Adaptive Temporal Modeling
Conference: 5th International Workshop on Event-Based Vision, CVPR 2025Link: https://arxiv.org/abs/2504.09960
Author(s): Hoang M. Truong, Vinh-Thuan Ly, Huy G. Tran, Thuan-Phat Nguyen, Tram T. Doan
Developed a custom data augmentation pipeline and contributed to the KnightPupil model, achieving a state-of-the-art Euclidean distance error of 1.61 on the 3ET+ benchmark through innovative spatiotemporal modeling and data augmentation techniques.
Projects
Some of my projects & contributions
Olympic Medal Prediction for MCM Competition
Developed data-driven models using Prophet time-series forecasting, KMeans clustering, and Randomized Inference to assess the impact of elite coaches on Olympic medal outcomes. Represented the University of Science, VNU-HCM in COMAP's Mathematical Contest in Modeling 2025.
Loading...
LLM-Powered Real Estate Valuation
Built a web scraper for batdongsan.com.vn, extracted 50k+ property listings, and used Gemini API for automated feature extraction from Vietnamese descriptions. Improved XGBoost model accuracy by 18% compared to manual processing for real estate price prediction.
Loading...
Event-Based Eye Tracking System
Developed a custom data augmentation pipeline and contributed to the KnightPupil model for real-time eye tracking. Achieved a state-of-the-art Euclidean distance error of 1.61 on the 3ET+ benchmark using PyTorch and spatiotemporal neural networks.
Loading...