Hi! I am Minh-Thien Nguyen, an AI Research Engineer specializing in Natural Language Processing (NLP) and Deep Learning.
My research interests include Embedding Models, Image-Text Retrieval for Vietnamese, Optimal Transport, Retrieval-Augmented Generation (RAG), and Image Classification. Additionally, I work on personal projects involving distributed training, TPU model training, and RAG architectures for complex document multiple-choice QA.
My publications (including preprints):
- ViCLIP-OT: The First Foundation Vision-Language Model for Vietnamese Image–Text Retrieval with Optimal Transport. [arXiv preprint]
- soups: Leveraging Model Soups to Classify Intangible Cultural Heritage Images from the Mekong Delta. [arXiv]
Some of my best projects include:
- seas: SEAS - A Smart Enrollment Advisory System for Can Tho University, built with async FastAPI, async SQLAlchemy, and async Qdrant.
- viettel-ai-race-vbkt: Multiple-Choice Question Answering (MCQA) Pipeline for Complex Technical Documents.
- medical-llama2: Med-Alpaca-2-7b-chat - A medical chatbot fine-tuned from the LLaMA 2 7B model.
- pre-training-gpt2: An end-to-end workflow for pre-training a GPT-2 model from scratch, with a focus on scalable training on XLA-enabled devices via PyTorch/XLA (CUDA and TPU).
Algorithmic Articles:
- Virtual tree/Cây ảo - VNOI Magazine, 2024 (magazine).
- Subtle Techniques with the Xor Operation/Kỹ thuật tinh tế về phép Xor - VNOI Magazine, 2023 (magazine).
- Read more in my hackmd blog.
Contact Information:
- Email: minhnguyent546@gmail.com
- X: @minhnguyent546
- Linkedin: /in/minhnguyent546