CONTACT

086*-***-***

adx8yt@r.postjobfree.com

linkedin.com/in/annguyenbk

Binh Thanh District, HCM City

EXPERIENCE

NEW OCEAN AUTOMATION SYSTEM - one of the leading automation solution providers for factories, machine builders and subcontractors in Vietnam and Southeast Asia.

06/2022 – 12/2022: Vision Intern

EDUCATION

HO CHI MINH CITY OF TECHNOLOGY

(BACH KHOA)

• 2019 - 2023

Control Engineering and Automation

Talent Engineer Program

GPA: 3.7/4.0

PROGRAMING

LANGUAGES

Python, C, C#, C++, Ladder, SQL, Assembly,

Schematic, Verilog.

KEY SKILLS

Optimize and develop AI/Deep learning

models

Learn about Machine Learning models and

some optimization algorithms

Operate Computer Vision and Robot systems

Implement AI models on linux or clouds

Annotate and preprocess data

Create and query the database system

Design and programming Embedded systems

& PLC

695 Toeic.

AWARDS

The author of a scientific research paper on AI –

BKCEEE 2023

2 Consolation prizes in National Physics Contest

for Seniors High School

PROJECTS

TRANSPARENT OBJECT 3D ESTIMATION

Optimize ClearGrasp - a Google's Deep Learning model by replacing sub-models and using Computer Vision algorithms to detect and determine the coordinates of transparent objects. ClearGrasp's sub-models are Semantic Segmentation, Boundary Detection and Surface Normal.

Use a stereo camera to capture images and a Stereo Matching model to create raw depth image for Depth Completion model whose output is a refined depth image. Then adopting YOLO-v7 to detect the center of transparent objects. Finnally, using a robot to grasp these things.

Implement models on Google Colab and use ngrok to connect to a GUI on latop. NEW DEPTH COMPLETION MODEL FOR TRANSPARENT

OBJECTS - CDTNET

Develop a new Depth Completion model - CDTNet, which's better than other state- of-the-art models on ClearGrasp dataset despite of being trained with less images. Combine DenseNet and Swin Transformer based on DFNet and TODE-Trans, also use the preprocessing of ClueDepth Grasp and ClearGrasp. BALL AND BEAM SYSTEM

Build the PID controllers so as to control and keep stable the position of the ball on a beam by microcontrollers STM32F103C8.

ANOTHER PROJECTS

Use MySQL to design and manage data of interaction of students and teachers. Train and test some object detection models such as YOLO v5-v7, EfficientNet, SSD and segmentation model like Unet, UNeXt, LWBNA Unet, MKDCNet, R-CNN. Some Data Engineering Projects on AirFlow, Azure and AWS. NGUYEN VAN AN

AI ENGINEERING