Curriculum Vitae · 2026
Tung Nguyen
AI Engineer / Researcher
Profile
AI researcher and engineer specializing in computer vision, generative artificial intelligence, and applied mathematics. Research and engineering work focuses on translating rigorous computational methods into reliable, production-ready systems, with experience in intelligent document processing, traffic analysis, retrieval-augmented generation, and speech enhancement.
Research & Experience
Full-time
AI Engineer
Viettel Solutions Corporation
- Research and develop intelligent document processing systems that extract structured information from forms, tables, scanned documents, and semi-structured layouts.
- Fine-tune vision-language models for document information extraction, including layout understanding, key-field recognition, and structured output generation.
- Develop layout-analysis and table-extraction modules that preserve reading order and support OCR, information extraction, and validation workflows.
- Build end-to-end license-plate recognition pipelines spanning image preprocessing, vehicle and plate detection, OCR, post-processing, and validation.
- Deploy license-plate recognition models with NVIDIA Triton Inference Server and optimize inference workflows for traffic-monitoring systems.
- Develop traffic-violation recognition features by combining detection, tracking, license-plate recognition, and rule-based video analytics.
- Research and develop speech-enhancement models for the Viettel Meeting platform.
Part-time
Research Consultant
WorldQuant BRAIN
- Research, design, and simulate quantitative financial alpha signals on the WorldQuant BRAIN platform using data-driven investment hypotheses.
Part-time
AI Engineer Intern
Grooo International JSC
- Researched and evaluated face detection, alignment, and recognition components for real-time computer-vision applications.
- Studied object-tracking algorithms under challenging conditions including occlusion, motion blur, and crowded scenes.
- Prepared datasets, ran model experiments, and integrated computer-vision modules into a camera-monitoring system.
- Evaluated performance and analyzed errors to improve model reliability, inference speed, and robustness.
Part-time
Data Scientist Intern
PIXTA Vietnam Co., Ltd.
- Researched facial-expression recognition and multitask-learning approaches for jointly learning facial attributes.
- Built experiments for face recognition, including dataset preparation, model training, evaluation, and performance comparison.
- Annotated, cleaned, and organized image datasets to improve training and validation data quality.
- Visualized datasets, label distributions, and predictions to support error analysis and model improvement.
Part-time
AI Engineer Intern
Techainer JSC
- Researched and implemented CenterNet-based object detection to localize Vietnamese identity cards in eKYC images.
- Developed card-alignment workflows to crop, rotate, and normalize identity-card images before text recognition.
- Applied OCR to extract identity information from Vietnamese identity cards.
- Tested the detection, alignment, and OCR pipeline across varied lighting, angles, and backgrounds.
Selected Publications
Selected Projects
P—01
dMOM Chatbot
A medical chatbot for the PHAD-sponsored dMOM project under Vietnam’s Ministry of Health. Fine-tuned large language models for Vietnamese medical question answering, built a retrieval-augmented generation pipeline grounded in trusted medical knowledge, and implemented a web application for interaction, demonstration, and evaluation.
P—02
AI Traffic Analysis
An end-to-end traffic-video analysis system combining vehicle detection and tracking, license-plate detection, and OCR. Trained and evaluated models on traffic datasets, integrated the modules into a unified backend pipeline, and built a web demo to visualize results.
View repositoryP—03
Vietnamese ID Card Recognition
An eKYC document-recognition pipeline for Vietnamese identity-card localization and alignment, with document cropping, perspective correction, image enhancement, text detection, OCR, and evaluation across varied capture conditions.