Thanh Duc Pham

I am an AI Resident at FPT AI Centre, working under the supervision of Prof. Pham Minh Tan. I received my Bachelor's degree from HUST (Hanoi University of Science and Technology). My research lies at the intersection of computer vision and deep learning, with a focus on efficient detection and segmentation architectures for real-world deployment.

I'm always open to collaborations, discussions, and new opportunities. Feel free to reach out if you're interested in my research or would like to discuss potential projects.

Research: My research centers on DETR-based object detection models for real-time applications, with particular emphasis on autonomous driving and remote sensing. I am also interested in frameworks related to the theoretical study of Neural Collapse in representation learning.

1. Efficient Detection & Segmentation. I develop transformer-based detection models (DETR variants) optimized for real-world settings such as self-driving systems and aerial imagery analysis, aiming to bridge the gap between model expressiveness and computational feasibility.

2. Knowledge Distillation. I investigate structured distillation strategies (Neural Collapse, Prompts) that preserve down-stream task representations while mitigating catastrophic forgetting in continual learning.