ML & Computer Vision
Model training, evaluation, and real-time inference pipelines.
AI Application Engineer · AI in Action @ VinUniversity
Building reliable systems — from REST APIs to ML pipelines. Applying AI engineering to production web/app systems, backed by research-driven projects.
Hanoi, Vietnam
Back-end developer with hands-on experience in RESTful APIs, database design, and production web systems for Japanese and Vietnamese clients — now focused on applying AI to real business problems, from computer vision and predictive models to AI-driven enterprise solutions, backed by award-winning research in AI/IoT and machine learning.
Currently studying AI in Action at VinUniversity and working as an AI Application Engineer — applying machine learning and generative AI to production web/app systems and databases, deepening skills in model deployment, cloud infrastructure, and intelligent systems while shipping clean, scalable software.
Model training, evaluation, and real-time inference pipelines.
RESTful services, business logic, and multi-tenant architectures.
Schema design, querying, and data pipelines for ML workloads.
Containerization, CI/CD, and production environment management.
Real-time fire detection combining YOLOv8 computer vision with IoT sensors. Web and mobile apps for monitoring, with continuous alerts via app and email.
ML pipeline to predict IT students' academic status — end-to-end from data collection and preprocessing to model serving via Flask API integrated with Laravel.
A three-layer knowledge graph — external regulations (SBV/Government circulars and decrees), SHB internal policies, and live banking products — modeled down to the Article/Clause/Point level. Automatically flags which products need re-review when an underlying regulation changes, and answers legal questions in Vietnamese with verifiable citations instead of plain document retrieval.
Hybrid BM25 + dense retrieval with cross-encoder reranking; every extracted legal relationship (amends/repeals/replaces/references) is verified before being trusted. Currently modeling 48K+ nodes and 167K+ edges across 757 fully-parsed legal documents.
Vinsmart Future · AI Chat Department
Applying AI to real-world web app workflows — building AI-powered features, chat-driven experiences, and large-scale web/app systems and databases for production products.
Built REST APIs, database schemas, and multi-tenant backends across Ohouse, Togismart, NEXS, Toshokan, and Job Portal projects for JP/VN clients.
TSO · Tinh Vân Software
ITC transport corporation website — requirements analysis and landing page development.
Currently building AI-driven products alongside large-scale web/app systems and databases — open to collaboration and technical partnerships.