Bùi Tùng Dương

Test Software Engineer at LG Innotek

Download Resume (PDF)

Professional Summary

Computer Vision and Software Engineer with 2+ years of full-time experience and one year of internship experience in camera systems, computer vision, edge AI, and manufacturing software.

Experienced in RAW Bayer processing, ISP preprocessing, image validation, algorithm integration, failure analysis, and high-volume production-data automation.

Built and validated image-processing pipelines against customer-provided reference implementations with numerical deviation below (10^{-9}), while delivering automation solutions that reduced validation and analysis effort by up to 82%.

Hands-on experience in AI model training, ONNX and TensorRT deployment, NVIDIA Jetson edge inference, camera calibration, geometric transformations, stereo vision, and real-time computer-vision systems.

Core Skills

Area Technologies and experience
Programming Python, C++, MATLAB
Computer vision and AI OpenCV, PyTorch, TensorFlow, YOLO, InsightFace
Camera systems RAW Bayer processing, ISP preprocessing, pinhole camera model, camera calibration, lens-distortion correction
Geometry and stereo vision Coordinate transformations, homography, pose estimation, stereo calibration, rectification, disparity, depth reconstruction
Edge AI NVIDIA Jetson Nano, ONNX, TensorRT, FP32, FP16, INT8
Tools and systems Git, PowerShell, manufacturing test systems
Engineering Algorithm integration, numerical validation, image processing, failure analysis, root-cause analysis
English TOEIC Speaking 140

Work Experience

LG Innotek Vietnam Haiphong

Test Software Engineer
January 2024 - Present · Haiphong, Vietnam

  • Developed and maintained production test software for main, ultra-wide, and telephoto smartphone camera modules using Python, C++, and MATLAB, supporting projects from NPI builds through PVT ramp.
  • Processed RAW Bayer image data and integrated customer-defined image-processing and camera functional-test algorithms into high-volume manufacturing test flows.
  • Reproduced and validated reference algorithms in production software, achieving numerical deviation below (10^{-9}) for verified outputs.
  • Developed image-processing and automated validation tools that reduced initial failure-analysis reporting time by 80%, validation cycle time by 60%, and manual analysis effort by 82%.
  • Investigated image and test-result anomalies by analyzing software behavior, preprocessing parameters, configuration, equipment conditions, and camera-module characteristics.
  • Communicated technical findings and root-cause conclusions with production, process, equipment, and customer teams, translating complex result differences into actionable corrective actions.

Universal Scientific Industrial Vietnam

SFIS Engineer Intern
July 2023 - December 2023 · Haiphong, Vietnam

  • Supported SFIS operations and investigated production-data synchronization issues across interconnected manufacturing systems.
  • Developed internal utilities to automate data processing and improve system-to-system data flow.
  • Automated SCP-based data synchronization during off-peak periods, replacing manual synchronization and reducing production waiting time.
  • Worked with manufacturing databases, system logs, and shop-floor applications to support production workflows and system integration.

VNNet Haiphong

Software Engineer Intern
January 2023 - June 2023 · Haiphong, Vietnam

  • Collected and prepared image data, trained a YOLOv5 fire-and-smoke detection model, converted it from PyTorch .pt to ONNX and TensorRT, and benchmarked FP32, FP16, and INT8 inference on NVIDIA Jetson Nano based on latency, throughput, and mAP.
  • Integrated automatic fire alerts through phone notifications and on-site alarm systems; the solution was accepted for deployment at a packaging manufacturing company in Hung Yen.
  • Developed a configurable face-recognition attendance application using Hanet AI-camera APIs for identity recognition, event handling, and attendance recording.

Selected Projects

Camera Geometry Toolkit

View source code

July 2026 - August 2026

A reproducible computer-vision toolkit covering pinhole-camera geometry, monocular calibration, stereo vision, and geometric transformations.

  • Implemented pinhole-camera projection, coordinate transformations, monocular calibration, lens-distortion correction, planar homography, and ArUco-based pose estimation.
  • Built a stereo-vision pipeline covering stereo calibration, image rectification, disparity estimation, and depth reconstruction.
  • Structured the Python implementation as reusable modules with runnable demos and validation scripts.
  • Developed OpenCV demos for environment verification and 3D-to-2D projection.

Open-Set Face Recognition

View source code

December 2021 - January 2022

An open-set face-identification application built with InsightFace.

  • Built a face-identification pipeline covering face detection, alignment, 512-dimensional embedding extraction, gallery enrollment, and cosine-similarity matching.
  • Implemented similarity-threshold rejection to classify identities outside the enrolled gallery as unknown.
  • Organized the application into reusable enrollment and inference workflows for adding and recognizing identities.

Global Wheat Detection

View project

July 2023

An object-detection project for identifying wheat heads in outdoor agricultural images.

  • Trained and evaluated an object-detection model for detecting wheat heads in outdoor agricultural images.
  • Implemented dataset preprocessing, augmentation, training, inference, and prediction visualization.
  • Analyzed false positives, missed detections, object-scale variation, and image-domain differences across datasets.

View more projects

Awards

LG Innotek Improvement Excellence Award 2026

March 2026

  • Developed a direct software interface between the production test system and automated visual-inspection equipment.
  • Accelerated inspection feedback and reduced the risk of defective units escaping downstream detection.
  • Delivered estimated annual cost savings of approximately KRW 600 million.

Data-Centric AI Competition 2021 — Prospective Prize

FPT Software · December 2021

  • Developed a face-mask wearing-condition detection pipeline for the COVID-19 context.
  • Improved data quality through augmentation, incorrect-label correction, bounding-box refinement, synthetic data generation, and model-assisted data collection.
  • Increased model performance from 0.64 to 0.91 mAP@0.5:0.95.
  • Ranked 2nd in the student leaderboard and 11th overall.

Education

Vietnam Maritime University

Bachelor of Information Technology
October 2020 - April 2024

  • GPA: 3.61/4.00

Contact