Dang Truc Dev
Third-year Computer Science student at VNU-UET, expected to graduate in June 2026. I have experience in AI/ML research and have built real-world systems powered by AI. I'm passionate about artificial intelligence and backend development. I enjoy turning smart ideas into scalable, production-ready products.
About Me
I have over a year of experience as a research assistant in AI/ML, and I’ve also contributed to the development of an interactive education platform aligned with Vietnam’s MOET curriculum. I'm passionate about both AI and backend development, as I enjoy building intelligent systems that are production-ready, efficient, and scalable. I’m always eager to learn from real-world projects where smart technologies meet clean, well-designed system architecture.
Dtruc in ISE Summer Trip 2025
My Journey
I enjoy coding every day especially backend stuff. Building smart, reliable systems that actually work makes me happy. Still learning, still loving it!
What Drives Me
I'm passionate about building technology that empowers people and even more passionate about breaking the myth that women don't belong in tech. I love backend development, enjoy mentoring aspiring devs, and stay curious about the latest in AI. Code speaks louder than stereotypes.
Data-Centric Development
Designing robust, reproducible pipelines with a focus on data quality, labeling, and preprocessing to drive better model performance
AI & Machine Learning
Working with deep learning models, LLMs, and multi-agent systems using frameworks like PyTorch, TensorFlow, and LangChain
Backend Engineering
Building clean, maintainable systems using Clean Architecture and design patterns — from REST APIs to scalable microservices
Technologies & Tools
Experience
My professional journey building innovative solutions
Intelligent Software Engineering Lab
Focused on data-centric research, including reading academic papers, proposing ideas, running experiments, and drafting publications
Key Achievements:
- Co-authored a Q1 research paper on data-centric AI (FGCS, 2025)
- First Prize - UET Scientific Research Conference for Undergraduates 2025
- First Prize - UET-FIT Scientific Research Conference for Undergraduates 2025
Thi De Education
Built an interactive E-learning platform aligned with the MOET 2025 curriculum, designed to enhance engagement and support personalized learning for high school students.
Key Achievements:
- Deployed a beta version used by 200+ students with a Dockerized infrastructure and NGINX reverse proxy.
- Leading migration to a cross-platform tech stack (Spring Boot, Golang, React, PostgreSQL, MinIO) to improve scalability, performance, and developer experience.
Auto Review Code Platform
Developed an automated code review tool tailored to teachers’ instructions, supporting scalable and efficient student code evaluation.
Key Achievements:
- Designed the backend using Clean Architecture with environment-based layering for smooth Docker deployment on servers.
- Implemented a queue system and gateway layer to handle chat requests with the system’s core Agent, preventing overload during peak student usage.
- Successfully deployed the MVP version in a production-like environment.
Featured Projects
Some of my recent work that I'm particularly proud of
AI-powered code review and grading system designed for educational institutions.
Automates student code evaluation using AI. Includes fully implemented RBAC. Courses and assignments can be migrated across subjects and reused. The system is pre-deployed and ready to use.
This product is ready for commercialization and can be hosted as a dedicated instance for individual clients. Contact me to request an admin account or to discuss deployment options.
E-learning platform aligned with the MOET 2025 curriculum, designed to enhance engagement and support personalized learning for high school students.
The system supports building a large scale question bank through real time document uploads and automatic OCR processing.
It’s ready for deployment and can be customized for institutional use. Contact me to request an admin account for the demo version.
Authstream - Lightweight, self-hosted authentication & authorization system
Co-designed a secure and high-performance auth system using Spring Boot and JWT, supporting both RBAC and ABAC models. Integrated real-time data synchronization between services via Debezium (CDC) and Redis Streams
Authstream is fully self-hosted and ready for integration into modern service-oriented architectures.
Automated UI Generation for ML Tasks
LoomUI is an end-to-end framework that automatically generates full-featured, visually consistent, and task-aware user interfaces for machine learning applications.
I'm open to collaboration and further development of this product. Contact me if you'd like to build on top of it.
Other Projects
Publications & Research
Academic contributions and research in AI, machine learning, and distributed systems
Featured Publications
Future Generation Computer Systems (Q1)
The performance of the Machine learning and Deep learning models heavily depends on the quality and quantity of the training data. However, real-world datasets often contain a considerable percentage of noisy labels, ranging from 8.0% to 38.5%. This could significantly reduce model accuracy. To address the problem of corrupted labels, we propose Cola, a novel data-centric approach that leverages both local neighborhood similarities and global relationships across the entire dataset to detect corrupted labels.