Data Science, Machine Learning, AI

Le Truong Giang

+ Hanoi, Vietnam # *****************@*****.*** +84-388-***-*** § itsZiang

About me

A recent Data Science graduate with expertise in data analysis, programming, and machine learning. Skilled in leveraging advanced technologies to solve complex problems and enhance project outcomes. I am seeking a full-time AI Engineer role where I can apply my skills in building and optimizing end-to-end AI systems. Education

Hanoi University of Science

BS in Data Science

Sept 2021 – July 2025

(expected)

GPA: 3.36/4.0

Honors & Awards: Academic Encouragement Scholarship (Semesters 3, 6) SAMSUNG Innovation Campus 2024

Learned about various concepts in mathematics, optimization, natural language processing (N-gram, Markov model, RNN, LSTM, Transformer), and computer vision (CNN, YOLO, DETR, U-net ) Experience

AI Engineer Intern

JVB Vietnam

Hanoi, Vietnam

June 2024 – Sept 2024

Owned the development of the OCR module and took part in building the pipeline for a real-time vehicle tracking and license plate recognition system.

Achieved 10–30 FPS on an RTX 2050 GPU and 87% accuracy for the license plate recognition task.

Technologies: YOLOv8, PaddleOCR, ByteTrack, OpenCV. Projects

Q&A chatbot for Vietnamese legal documents (Graduation thesis)

Developed a Retrieval-Augmented Generation (RAG) chatbot for Vietnamese traffic laws, implementing an advanced retrieval pipeline (hybrid search, reranking, Google Search) to achieve high accuracy.

Achieved performance on a golden dataset: 92.1% Recall@3, 86% MRR@3 (+7.8% by fine-tuning the reranker model), and a 4.7/5 correctness score evaluated by GPT-4.1-mini.

Technologies: Qdrant, MariaDB, Google CSE, GPT-4.1-mini, Qwen 3 8B, BGE-M3, Streamlit. Aspect-based Sentiment Analysis for Vietnamese Reviews on E-commerce Websites

Built a scheduled crawler pipeline for automatically crawling user reviews about electronic device products.

Fine-tuned an Electra model for Aspect Detection and PhoBERT for Sentiment Analysis, achieving 77% F1- score and 89% F1-score for Aspect Detection and Sentiment Analysis, respectively.

Technologies: PhoBERT, Electra, Shopee API.

Hate Speech Detection on YouTube

Developed a YouTube hate speech detection tool (YouTube Data API) with features for automated moderation and comment analytics.

Fine-tuned the PhoBERT model for a three-class classification task, improved F1-macro from 61% to 70% and accuracy from 80% to 88% through data augmentation and data preprocessing techniques.

Technologies: PhoBERT, MongoDB, YouTube Data API. Skills

Training machine learning and deep learning models using Pytorch

Deploying services with deep learning models using FastAPI

Utilizing relational databases (MySQL) and vector databases (Qdrant)

Language: Vietnamese (native), English (TOEIC 800)

Contact this candidate