VIESGVIESG
Back to VIESG
H

Tran Thi Ngoc Han

Computer Vision Engineer · VIESG

han@viesg.com+84 976 986 124Ho Chi Minh City, Vietnam
Download CV

Career Objective

Data Science student with a passion for Computer Vision, Deep Learning, and quantization techniques on Edge/Mobile devices. Experienced in academic research with accepted papers at international conferences and AI solution competitions. Focused on deploying efficient AI models on edge devices through quantization and mixed-precision techniques.

Research Experience

Researcher & Technical Writer — Multi-modal Assistive Device

10/2025 – Present

HUFLIT, HCMC, Vietnam

Bidirectional, multi-modal communication device (voice + text + gesture) operating offline for people with disabilities. Demonstrates viability for deployment in healthcare, education, and care settings in Vietnam.

Multi-modal AIOffline DeploymentAccessibility

Researcher & Technical Writer — Hydrogen Leakage Detection System

03/2025 – 04/2026

HUFLIT, HCMC, Vietnam

System using thermal cameras and hydrogen sensors to detect hydrogen leaks on Raspberry Pi 5 based on the Joule-Thomson effect. Integrates alert horn and email notifications; deployed in containerized environments.

Thermal VisionRaspberry Pi 5IoTPython

Developer — Offline Election Information Chatbot

2025

01 Ly Thai To, Water Drop Park, HCMC

Contributed to the offline AI chatbot for the 'Học sinh, sinh viên Thành phố Hồ Chí trao niềm tin, gửi kì vọng' community event providing election information to the public.

NLPOffline AIChatbot

Developer — Footwear Defect Detection on Conveyor Belt

2025 (In Progress)

HCMC, Vietnam

Building an Android-based edge device to detect footwear defects on an industrial conveyor belt and report findings to a server. Demonstrates a portable yet powerful approach to industrial inspection.

AndroidEdge AIComputer VisionIndustrial

Publications

1st AuthorATAC 2026Best Paper Award

Real-Time Traffic Sign Detection for Edge Devices: A Mixed-Precision NMS-Free Approach

The 2026 Conference on Applications of Technology, Automation and Civil Engineering

Proposed NMS-free YOLO26 framework — 46× model compression (2.85 MB), 27× speed-up on mobile while preserving 99% original accuracy.

Co-authorMLCI 2026

A Multi-Modal Approach for Hydrogen Leakage Monitoring based on Thermal Vision

2026 International Conference on Machine Learning and Computational Intelligence

Thermal camera + hydrogen sensor fusion on Raspberry Pi 5.

Co-authorMLCI 2026

An Android-Based Edge Device for Footwear Defect Inspection Toward a Portable yet Powerful Approach

The 2026 International Conference on Machine Learning and Computational Intelligence

Education

HUFLIT

Ho Chi Minh City University of Foreign Languages — Information Technology

B.Sc. Information Technology

Major: Data Science

2023 – Present

GPA: 3.18

New Student Scholarship

Awards

Top 9 — UI Challenge

Bach Khoa Innovation 2025

Consolation Prize

IT Got Talent — Hutech 2025

Skills

Deep Learning

YOLO Family (YOLOv5–YOLOv8, YOLO26)CNNLSTMTransformerLLM

Edge Deployment

ONNXTensorRTINT8/FP16 QuantizationQATMixed-Precision

Hardware

Android Edge DevicesESP32ArduinoRaspberry Pi 5Orange Pi 5

Languages & Tools

.NET (C#, ASP.NET)Python (PyTorch, OpenCV, TensorFlow)C++ (Arduino)GitHubSQL ServerMongoDB