Information Technology Machine Learning

NGUYEN TAN DUNG

+843******** # ************************@*****.*** § github.com/Dunglqd ï linkedin.com/in/tandunglqd Education

VNUHCM - University of Information Technology (UIT) Expected May 2025 Bachelor Degree in High-Quality Information Technology Programs (GPA: 3.27 / 4.00) Ho Chi Minh

• IELTS: 5.5

• JLPT: N3

Projects

Chatbot system for UIT Admissions ChromaDB, LLM, Generative AI, Prompt Engineering, Llama-index, API, MongoDB, BM25

• Leverage RAG (Retrieval-Augmented Generation) with hybrid search (Vector and Keyword) to gather and classify admission information for UIT.

• Design prompts for the generation module and integrate modules such as retrieval, classification, and reranking to provide accurate responses.

• Implement a vector database with ChromaDB and additional databases with MongoDB.

• Build the user interface using Streamlit and connect the backend via FastAPI.

• Technologies used: Llama-index, Langchain, MongoDB, ChromaDB, Redis, FastAPI, Pydantic, Streamlit, Chainlit, Transformers, Torch, Sentence Transformers, Selenium, BeautifulSoup4, Pandas, Underthesea, VNCoreNLP, Lingua DDoS Attack Detection System Machine Learning, Kafka, Spark Streaming, TensorFlow

• Leverage Machine Learning with hybrid models (XGBoost, RBF-SVM, Decision Tree, and DNN) to detect DDoS attacks using Kafka streaming data.

• Design and implement real-time data processing and feature extraction using Apache Kafka and Spark Streaming.

• Deploy models using TensorFlow/Keras for analyzing network traffic and predicting DDoS attacks.

• Evaluate the performance of each model with metrics like accuracy, precision, recall, and F1 score, achieving up to 99.18

Vietnamese Sentiment Analysis System ChatGPT, BERT, ViBERT, SVM

• Integrate ChatGPT for Vietnamese sentiment analysis alongside traditional models like BERT, ViBERT, and SVM.

• Process and classify Vietnamese text data using natural language processing techniques.

• Evaluate model performance with metrics like accuracy, precision, recall, and F1 score, with notable improvements in sentiment prediction accuracy.

• Implement feature extraction and text preprocessing methods tailored for Vietnamese language specifics.

• Develop a pipeline combining machine learning and deep learning models for robust sentiment analysis of Vietnamese text data.

Real-time Toxic Comment Analysis From Live YouTube Videos Pytchat, Kafka, Apache Spark, PhoBERT, Docker

• Build an application for analyzing sentiment of toxic comments on live YouTube videos.

• Use Pytchat for comment extraction, Kafka and Spark for data processing, PhoBERT for classification.

• Train the model on the ViHSD dataset (Vietnamese Hate Speech Detection).

• Achieve 85.34 percent accuracy in comment classification.

• Develop a web application to visualize the analysis results. Hotel Management System SQL, C, 3-Layer Architecture

• Develop hotel management software with basic features: booking, check-out, reporting.

• Define business requirements using forms and regulations, then model them with data flow diagrams.

• Design the system with 3-layer architecture, database, and user interface, then implement and test core functionalities.

Achievement

• Top 5 in Bosch Coderace Challenge 2023 - Exceed the Limitless Mind.

• Certificate ”Sinh Vien 5 Tot” by The University of Information Technology in 2022, 2023 Publication

• Exploring the Performance of ChatGPT for Vietnamese Sentiment Analysis, UIT Young Scientists and Fellows Conference, 2023. Dai Nguyen Ba, Nguyen Tan Dung, Dang Van Thin. Technical Skills

Generative AI • Prompt Engineering • API • Text Mining • Tensorflow • Pytorch • Problem Solving • NLP • Git and Github • Docker • MLOps • SQL

Contact this candidate