I am a creative data scientist with five years' experience in the industry. I have experience in designing and carrying out experiments, testing, researching scientific literature, and writing technical reports. I possess excellent analytical and communications skills and a dedicated approach to working in a highly controlled working environment.

Programming Languages: Python/C++  

Skills: 

• Python, C++, and CUDA.  

• Linux, Docker

• Milvus DB

• Numpy, Pandas, Openpyxl.

• PaddlePaddle, Pytorch, DGL, Scikit-learn, ONNX, OpenCV

• Computer Vision, Deep Learning, Computer Graphic, Business Intelligence 

Tools: Pycharm, Jupyter, Visual Studio Code, Git, Tableau, Power BI

Auto License Plate Recognition System – UK Client

   Developer, May 2024 – Present

Project description 

• Develop an Automatic License Plate Recognition (ALPR) system to enhance vehicle identification and management using advanced computer vision and machine learning techniques.

Responsibilities 

• Researched and selected optimal techniques for license plate detection and recognition; fine-tuned pretrained models to minimize data labelling.

• Designed project requirements, developed tools for data loading, preprocessing, and augmentation, and monitored training progress.

• Evaluated and compared ALPR techniques, generating performance reports and selecting the most effective solutions.

• Converted models to ONNX format for client system integration and created API endpoints for testing and validation.

• Deployed models, ensuring seamless integration and performance in client environments.

Technologies 

• Python, OpenCV, PyTorch, ONNX, PaddlePaddle, Flask

‍

Invoice Extractor – UK Client

Developer, January 2024 – May 2024

Project description 

• This project is built to create an OCR invoice extractor for clients to optimize the documentation processes.

Responsibilities 

• Investigated and assessed OCR process options, selected the most effective solution, and utilized pretrained models to reduce labelling costs.

• Designed project requirement estimation plans and built tools for data loading, model training, progress monitoring, and success measurement in text detection.

• Created visual representations and reports to showcase image transformations, model performance, and other key details for clients, converting OCR models to ONNX format for system integration.

• Surveyed solutions for key information extraction, compared performance, and processed data from bounding boxes to graph data for training Graph Convolution models.

• Deployed models and created endpoints for team testing, ensuring seamless integration and performance.

Technologies 

• Python, OpenCV, PyTorch, ONNX, PaddlePaddle, dgl - Deep Graph Library, Flask.

‍‍

Biometric Attendance System – UK Client

Developer, July 2023 – December 2023

Project description 

• This project is built to upgrade the biometric system of the company.

Responsibilities

• Researched, surveyed, and evaluated solutions for face detection (SCRFD, ULFD) and face verification (ArcFace, MagFace); analysed and pre-processed data for training and evaluating models, including extreme conditions simulations.

• Collaborated with Backend and Frontend teams to integrate facial models and vector databases (Faiss, Milvus), built Docker implementations for client servers, and supported clients in JavaScript and .NET environments.

• Developed tools for comparing input and output data, ensuring model accuracy and performance, and created modules for simulating face images in extreme conditions.

• Researched, collected, and processed datasets for face anti-spoofing; decoded and reconstructed PyTorch models from ONNX; integrated domain generalization training; and planned the collection of private datasets.

• Trained and evaluated face verification and anti-spoofing models, ensuring robust performance in various conditions.

Technologies

• Python, C#, ReactJS, Pandas, OpenCV, PyTorch, ONNX, Milvus, Docker.

‍

Sale Performance Analysis – Europe

BI, March 2023– July 2023

Project description 

• This project analyses sales performance products from a TV entertainment show.

Responsibilities

• Migrated Looker Studio reports to PowerBI, ensuring data accuracy and visual appeal.

• Validated data in BigQuery and PowerBI using Python, Pandas, and Numpy, resolving discrepancies for enhanced data quality.

• Interpreted client needs, formulated custom sales metric equations, and delivered actionable insights.

• Established and managed complex table relationships (Star schema) for efficient data organization.

• Applied advanced statistics and data transformation to refine raw data for better analysis and designed dynamic reports with customizable filters using DAX formulas to exceed client expectations.

Technologies

• Python, Pandas, SQL Database, Power BI, DAX, Looker Studio, BigQuery  

‍‍

Gold Price – Vietnam

Developer - BI, February 2023– March 2023

Project description 

• This project analyses gold price data for our client.

Responsibilities

• Clean and research data from Data Lake.

• Developed an interactive Looker Studio dashboard displaying daily, weekly, and monthly gold price trends.

• Created a precise machine learning model for real-time gold price prediction.

• Crafted customized Reports and Dashboards in Looker Studio to meet client requirements.

Technologies

• Looker Studio, Pandas, SQL Database, Openpyxl.

‍

Marketing Performance Analysis – Vietnam

BI, January 2022 – February 2023

Project description 

• This project analyses marketing investing - revenue data for our client.

Responsibilities

• Clean and research data from sales reports.

• Generate report, dashboard to visualize marketing spend and revenue ratio for analysing to save marketing fees by 20%.

• Generate dashboards, use regression methods to find new potential locations.

• Build Reports and Dashboards using Tableau follow client requirements.

Technologies

• Tableau, Pandas, SQL Database, Openpyxl.

‍

Face Attendance – Vietnam

Developer, October 2022 – January 2022

Project description

• R&D Deep learning models and Image processing algorithm for face detection, tracking and recognition.

Responsibilities

• Developed baseline modules and optimized pipelines for face detection and tracking on edge devices.

• Created Image Processing algorithms for face tracking and a 3D-mask generator module for masked face adaptation.

• Built APIs for face recognition server, including adding members and stress testing modules to estimate inference time.

• Researched, trained, and evaluated face recognition models to enhance performance.

• Implemented message queue systems and created Docker images for packaging.

Technologies

• 3D ,Pytorch, OpenCV, ONNX, Python, Kafka, SocketIO.

‍‍

AI-CCTV – Vietnam

Developer, October 2022 – December 2022

Project description

• R&D Deep learning models and Image processing algorithm for CCTV application.

Responsibilities

• Research and optimize YOLO for human detection.

• Collect, analyze and rebuild dataset for human detection and anomaly detection.

• Develop a baseline for modules related to human detection and tracking.

• Estimate hardware stat for AI system to meet Real Time requirement.

Technologies

• Pytorch, OpenCV, ONNX, Python

‍

eKYC – Vietnam

Developer, October 2022 – December 2022

Project description

• R&D Deep learning models and Image processing algorithm for Error domain classification.

Responsibilities

• Develop a baseline for modules related to image quality classification.

• Build a tool to simulate and generate synthetic data that resembles the error domain of OCR model.

• Analyze, define failure cases, generate reports and continue to improve AI models.

Technologies

• Pytorch, OpenCV, Python, Pandas, GradCAM

‍

Beauty App – Vietnam

Developer, August 2021 – September 2022

Project description

• R&D Deep learning models and Image processing algorithm for building beauty apps.

Responsibilities

• Developed baseline modules for face segmentation, GAN for face makeup, and 3D face modelling.

• Collected, analysed, augmented, and generated data for training and testing.

• Trained, fine-tuned, and improved models to meet specific requirements.

• Researched and implemented solutions for integrating AI models into production, including model conversion and optimization (TensorRT, ONNX).

• Developed and integrated Image Processing and Deep Learning models, including C++ implementations for OBS Plugin (Live Streaming open-source software).

Technologies

• Pytorch, OpenCV, Python, Pandas, GradCAM.

‍

Image Denoise – Vietnam

Developer, June 2020 – December 2020

Project description

• R&D Deep learning models for Gray Image Denoising

Responsibilities

• Research and retrain State of the art deep learning model.

• Create new images denoising the deep learning models by applying deep learning methods.

• Training and evaluating image denoising deep learning models.

Technologies

• Pytorch, Keras, Python

‍

Auto Parking – Vietnam

Developer, January 2020 – April 2020

Project description

• Building an app to manage motorbikes in an indoor park.

Responsibilities

• Research and develop a pipeline to read motorcycle license plates by combining image processing and machine learning.

• Manage the number of motorbikes in the park with pandas.

Technologies

• OpenCV, Pandas, Python

Natural Language Processing with Classification and Vector Spaces

Awards:

• Second prize in the Student Scientific Research Competition of the Faculty of Electronics & Telecommunication Engineering, for the paper “A Flexible Residual Neural Network for Rician Noise Removal in MR Images,” presented at the 2024 Tenth International Conference on Communications and Electronics (ICCE), Danang, Vietnam.