Deep Learning Artificial Intelligence

Hoang Phi Hung

AI Intern

085******* ****************@*****.*** github.com/hoangphihung2004 Da Nang, Viet Nam Objective

My goal is to take advantage of internship opportunities to develop practical skills and subject matter expertise in my field, especially in Artificial Intelligence and Machine Learning. I want to be part of a dynamic and innovative organization where I can apply my knowledge and learn from experienced people. Connectivity, willingness to learn, and a passion for AI are my strengths, and I look forward to being able to contribute positively to the organization’s goals and mission.

Education

FPT University Da Nang 2022 – 2026

• Artificial Intelligence (AI) - GPA: 3.75/4.0

Honors & Awards

Finalist of the RESEARCH FESTIVAL 2025 National Scientific Research Competition.

Jul 2025

Champion of the RESEARCH CONNECT 2025 Scientific Research Competition. Jun 2025 Projects

Developing an Intelligent Q&A System for Vietnamese History using RAG (Retrieval-Augmented Generation)

• Built a dual-function Q&A system that enables users to ask questions about Vietnamese history using either a curated knowledge base or their own uploaded PDF documents.

• For historical data, employed a hybrid retrieval pipeline with ElasticSearch (BM25) and Qdrant for semantic search, combined with large language models (e.g., Gemini) to generate accurate, cited responses.

• For PDF documents, parsed and chunked uploaded files, then embedded them using Transformer models and stored in FAISS (via LangChain) for in-memory semantic retrieval and context-aware answering. Note: Source code available at https://github.com/hoangphihung2004/Viet_Su_Tri_VietNamese_History. Optimizing a Rice Classification Model for Real-time Applications using Knowledge Distillation

• This study explores the application of Knowledge Distillation (KD) techniques for rice variety classification. A lightweight student model is trained under the guidance of a more complex teacher network to achieve high accuracy while reducing computational cost and significantly lowering the number of model parameters.

• The approach achieves better results than standard pre-trained models and also this makes it suitable for deploy- ment on resource-constrained devices such as mobile phones or embedded systems. Enhancing Potato Leaf Disease Classification through Handcrafted Feature Engineering

• Focused on the classification of potato leaf diseases from images captured under challenging, real-world field conditions.

• Focused on feature engineering by extracting and combining color and texture features (SIFT, KAZE, BoVW) from images. Applied Borderline-SMOTE to handle data imbalance, ensuring the model learned equally across all disease types. Trained and fine-tuned classical machine learning models to solve an image classification task. Leveraging Handcrafted and Deep Features for Automated Fish Freshness Assessment

• Conducted comprehensive research and evaluation of computer vision approaches for automating fish freshness classification based on eye images. Extracted and analyzed a comprehensive set of handcrafted visual features, including texture (GLCM, LBP) and multi-space color statistics. Fine-tuned and evaluated multiple state-of-the- art deep learning architectures for the classification task.

• Proposed a novel hybrid pipeline that combines handcrafted features with deep learning features, optimized using LightGBM-based feature selection. Used Grad-CAM to visualize and validate which image regions the deep learning models focused on during decision-making. Note: Currently preparing a manuscript on these studies for submission to journals and conferences, so it cannot be made public on GitHub at this stage — but I am happy to share and discuss it in interviews. Skills

English

• Proficient in reading English technical documents, with basic communication skills. Technologies

• Languages: Python

• ML/DL Frameworks: PyTorch, Tensorflow, Scikit-learn

• Data Science & CV: Pandas, Numpy, Matplotlib, Seaborn, OpenCV (cv2)

• LLM & RAG Technologies: LangChain, FAISS, Qdrant, ElasticSearch

• Databases: SQL Server

• Tools & Version Control: Git

Contact this candidate