cv

Master

• Master of Science in Computer Vision.

Bachelor

• Bachelor of Science in Data Science and Artificial Intelligence.
• 3.96/4 GPA, salutatorian.
• Thesis: Mamba-based Constrastive Language-Image Pretraining for Endoscopic Image Analysis – 9/10.
• Relevant courses: Calculus, Algebra, Probability & Statistics, Machine Learning, Data Science, Deep Learning, Computer Vision.

Research Assistant

• Member of Applied AI in healthcare group (collaboration with AIML - University of Adelaide).
• Current project: Hallucination in Medical Visual Question Answering.

Research Assistant

• Member of Computer Vision research team.
• Teaching Assistant in Introduction to Deep Learning course in semester 2023.1.

Localizing Before Answering: A Benchmark for Grounded Medical Visual Question Answering (IJCAI 2025)

• To improve visual reasoning, we propose the Localize-before-Answer (LobA) framework, which trains LMMs to localize target regions of interest and self-prompt to emphasize segmented pathological areas, generating grounded and reliable answers.

TrafficVLM: A controllable visual language model for traffic video captioning (CVPR 2024 AI City Challenge)

• TrafficVLM models traffic video events at different levels of analysis both spatially and temporally and generates long fine-grained descriptions for the vehicle and pedestrian at different phases of the event.

Software & Data Engineer

• Integrate customers’ data from old storage to new system on S3 using Apache NiFi and use ffmpeg, opencv and PIL to validate media integrity after transfer.
• Integrated data from internal management systems such as Jira to evaluate employee performance using ETL Pentaho and Apache NiFi, helping decrease workload for HR department by 80%.
• Built a system to conduct internal surveys for 1500 employees and provide biannual rating reports automatically on Oracle APEX with modifications of JavaScript, CSS and storage on Oracle SQL Server, reducing a 15 man-days task to 0.5 man-days.

AI Researcher Intern

• Researched segmentation-based models derived from Unet++ and DNA-Net; object tracking method based on SORT and ensemble two models to improve detection accuracy for the IRST (InfraRed Search and Track) System by 5% compared to non-deep learning methods.
• Implemented models using Python libraries (PyTorch, OpenCV); serving with ONNX and TensorRT.

• Australia-Vietnam Strategic Technologies Scholarship
  • Awarded by Australia-Vietnam Strategic Technologies Centre, University of Technology Sydney
• Scholarship for STEM disciplines
  • Awarded by German Academic Exchange Service (DAAD)
• Padova Excellence Scholarship
  • Awarded by University of Padua
• Rising Scholars Fellowship
  • Awarded by Singapore Management University

• Certificate for Excellent Graduate
  • Awarded by Hanoi University of Science and Technology

• Exness Scholarship
  • Awarded by Exness Scholarship

• 6x HUST Excellent Scholarship
  • Awarded by Hanoi University of Science and Technology