Deep Learning Machine

+849******** · ********.******@*****.*** · www.linkedin.com/in/tamnguyentrong-uit

Tan Binh District, Ho Chi Minh City

Trong-Tam Nguyen

I am an individual full of ambition and passionate about Artificial Intelligence . My aspiration is to attain expertise in AI, with a focused aim to advance further in the field of artificial intelligence research over the next 2-3 years. With a solid educational background in Computer Networks and Data Communications, I am dedicated to bringing my utmost commitment to the table and forging a long-term partnership with the company, should the opportunity arise. My unwavering enthusiasm, resolute determination, and steadfast dedication to personal and professional growth aligned with the company's objectives are attributes I hope prospective employers will duly consider. OBJECTIVE STATEMENT

Majors: Computer Networks and Data Communications GPA: 3.2/4 TOIEC-LR: 665 (2022) EDUCATION

University of Information Technology - HCMC VNU August 2021 - Present KEY COMPETENCIES

Green Summer Volunteer Campaign

Thu Duc, Ho Chi Minh City

CERTIFICATIONS EXTRACURRICULAR ACTIVITIES

Machine Learning Specialization

Coursera

Deep Learning Specialization

Coursera

Critical Thinking

Thinking School

AIO2023

ML/DL

COURSES

Programming Language

C++, Python, SQL, Statistics

Machine Learning Algorithms

Deep understanding in both

supervised and unsupervised

learning

Mathematics and Statistics

Solid foundation in probability, linear

algebra, and calculus.

Deep learning

Experience with deep learning models

like CNNs, RNNs, Transformer

Microsoft Office

Word, Excel, Powerpoint

Library and framework

Familiar with sci-kit learn, numpy,

pandas.

Experience with Pytorch, Tensorflow

A I E N G I N E E R

Deep learning for

Computer Vision-

Stanford 2017

PROJECT 1: UTILIZING DEEP LEARNING FOR HEART RATE MONITORING USING THE MAX30102 SENSOR. Description: Developing a health monitoring system using a heart rate monitoring device and applying deep learning for heart disease prediction.

Methods:

Preprocessing for tabular data

Building a database that allows real-time updating of patient data during the monitoring process Buid and train Neural network model using Tensorflow Deploy model on app to predict heart disease

Achievements:

A model with an accuracy of over 88% on the training set and over 83% on the test set. A system capable of predicting heart disease based on heart rate data. A model with an accuracy of over 88% on the training set and over 80% on the test set. Link Paper

ACADEMIC PROJECTS

PROJECT 2: YOLOV8 FOR AMERICA LANGUAGE TRANSLATION Project Description: Developing a system capable of recognizing and translating sign language images into text from video inputs, and deploying the model as a service (API) using FastAPI. Methods:

Preparing data: Developing a system capable of recognizing and translating sign language images into text from video inputs, and deploying the model as a service (API) using FastAPI. Finetuning model: Re-trained the pre-trained YOLOv8n model with adjusted learning rate and batch size using the AdamW optimizer. Freezing the first ten layers expedited the training process. Evaluation Parameters: Measured effectiveness using mAP(Mean Average Pricision) Achievements:

The YOLOv8 model exhibited superior performance The model achieved the highest mAP of approximately 0.9 on the training set at IoU (Intersection over Union) threshold of 0.5, and nearly 0.8 mAP50-95 on the test set

Depoly model as service (API) using FastAPI with video inputs Link Paper

PROJECT 4: ABNORMALY DETECTION ON X-RAY IMAGES (IN PROGRESS) Project Description: Using different object detection models such as DETR, YOLOv10 for detecting lesions on chest X-ray images, along with additional image preprocessing, data augmentation techniques to enhance the performance of the model.

Methods:

Utilizing image preprocessing techniques such as blurring, denoising, thresholding, etc. Synthesizing evaluation results based on combining various models with different preprocessing and augmentating techniques.

Evaluation Parameters: Measured effectiveness using mAP Achievements:

ADDITIONAL SKILLS

Information Elicitation & Requirement Analysis

Negotiation & Problem Solving

Documentation and Read paper

Data Preprocessing, Data Cleansing, Data Integration Communication skills

Good at Team work

Attention to detail

Data visualization

Natural Language Processing (basic) & Computer Vision PROJECT 3: SEGMENTATION TASK ON KVASIR-SEG DATASET Description: This paper focuses on the task of polyp segmentation in medical imaging, which is crucial for the early detection and treatment of colorectal cancer. Methods:

Image Preprocessing: Applied methods such as specular highlight removal to improve image quality. Models: Used CNN models including SegNet, DuckNet, U-Net, and various U-Net variants. Evaluation: Employed metrics such as Dice coefficient, Intersection over Union (IoU), Sensitivity, and Specificity to assess segmentation accuracy.

Achievements:

DuckNet and U-Net variants outperformed other models in accuracy and robustness. Traditional U-Net and SegNet did not perform as well, indicating the necessity for more advanced architectures and preprocessing techniques

Link Paper

Contact this candidate