Thong Truong
Full Stack AI Engineer
I build intelligent, scalable full-stack applications with expertise in AI integration, containerized deployment, and high-performance systems.
Creator of Testli AI and specialist in AI-powered applications with scalable architectures
Frontend Architecture
- Next.js App Router & SSR
- TypeScript & React Best Practices
- Angular & RxJS Reactive Programming
- WebSocket for Real-time Updates
Backend Systems
- FastAPI, Async ORMs & Pydantic
- AI Integration (LLMs, RAG)
- NestJS & TypeScript Microservices
- Spring Boot & Java Enterprise
Deployment & DevOps
- Dokploy (Self-hosted Deployment)
- Docker & Compose Architecture
- CI/CD with GitHub Actions
Full Stack Projects
Testli AI - AI-Powered English Exam Generator for Vietnam's National High School Graduation Exam
A next-generation platform that leverages AI to automatically generate and personalize English exam questions, fully aligned with the official format of Vietnam's National High School Graduation Exam. Enjoy real-time progress feedback and a seamless, modern user experience.
Demo: app.testliai.com →Frontend Highlights
- • Built with Next.js and shadcn/ui for a responsive, accessible interface
- • TanStack Query for efficient data fetching and caching
- • Real-time task progress and exam updates via WebSocket integration
Backend Architecture
- • FastAPI with Swagger/OpenAPI for robust, well-documented APIs
- • Gemini GenAI for intelligent exam generation, featuring context caching and structured outputs
- • High-performance database operations using SQLAlchemy async ORM with Alembic migrations
- • Redis for high-performance caching and real-time task tracking
- • Dramatiq for scalable background job processing
Key Achievements
- • End-to-end exam generation in under 10 seconds
- • Smooth, uninterrupted user experience through efficient background task delegation
- • Real-time task and progress updates delivered via WebSocket
- • Strict adherence to Vietnam's National High School Graduation Exam standards
System Architecture
SaaS Chatbot Platform for Universities & Student Support
An AI-powered multi-tenant chatbot system designed for educational institutions. This platform helps universities deploy contextual chatbots on their portals to enhance student learning, answer questions, and deliver support—at scale.
Each university acts as a tenant, with isolated access to their institutional knowledge base. Students log in via SSO, and the chatbot personalizes responses using both global university data and student-specific uploaded materials. Through hybrid search (semantic + keyword), the system delivers precise, source-linked answers using Retrieval-Augmented Generation (RAG).
Frontend Highlights
- • Next.js + TypeScript for admin dashboard
- • Embeddable React widget with Web Components
- • Real-time chat UI with WebSocket integration
- • SSO-compatible integration for student login
Backend Architecture
- • FastAPI microservices with tenant-aware JWT authentication
- • PostgreSQL with row-level security for per-tenant and per-user isolation
- • Hybrid search: Vector DB (Milvus/Pinecone) + Elasticsearch
- • RAG flow powered by OpenAI or other LLMs
- • Redis used for caching and real-time chat messaging
Key Capabilities for Universities
- • Student SSO login via university auth systems
- • Client & user dashboards to upload documents, notes, links
- • Crawl4AI for real-time web page ingestion
- • Automatic indexing of content (text, PDFs, links)
- • Chat history and source-aware responses
- • Usage analytics & conversation logs
- • Stripe-based billing for flexible institution plans
System Architecture
Academic Projects
Selected projects from my university coursework demonstrating my expertise in AI, data science, and software development.
BGU118 - Financial Asset Tracker
A Python web application integrating established financial APIs to track and analyze Bitcoin, gold, and USD prices with real-time visualization.
Technologies
Key Features
- •Integrated multiple financial market APIs for reliable data sourcing
- •Implemented automated data aggregation and normalization pipeline
- •Developed responsive dashboards with real-time price updates
Bookstore Recommendation System
A collaborative and content-based recommendation system for a bookstore, providing personalized book suggestions.
Technologies
Key Features
- •Implemented collaborative filtering algorithms
- •Developed content-based recommendation features
- •Optimized for performance with large datasets
Movie Recommendation System
Contributed to a social network project that implemented a movie recommendation system using LightGCN for collaborative filtering.
Technologies
Key Features
- •Implemented LightGCN architecture for collaborative filtering
- •Worked on graph-based recommendation algorithms
- •Contributed to model training and evaluation processes
Pet Clinic Management System
A Java Swing application with JDBC for veterinary clinic management, following MVC architecture and DAO pattern.
Technologies
Key Features
- •Implemented MVC architecture with DAO pattern for data access
- •Built responsive user interface with Java Swing
- •Developed MySQL database integration via JDBC
- •Created comprehensive clinic management features
Let's Build Something Amazing
Looking for a full-stack developer who can architect and implement complete solutions? Let's discuss your project.