Nguyen Dat

Data Science final year student and AI researcher with experience in Natural Language Processing, Vision-Language Models, and formal planning. Passionate about building training-free reasoning systems that bridge visual perception and symbolic decision-making, with current research focusing on multimodal planning, Large Language Models, and neural-symbolic reasoning.

Machine Learning Engineer at NLP@UIT (2024-06 – 2025-10)

• Led a team of 3 to research and re-implement language models, including RNN and Transformer - based, for long-context abstractive text summarization, improving ROUGE scores by 10% compare to original papers.
• Designed and implemented MMF, a modular deep learning framework in PyTorch, supporting multiple model architectures and flexible data pipelines, enabling scalable experimentation across 6+ models and 3+ dataloader strategies.
• Developed a Vietnamese phoneme-based model, mitigate long dependency and solving OOV using phoneme in language model.

AI Research Intern at AIO (2026-01 – Present)

• Conducted research on training-free Vision-Language Model (VLM) planning systems, focusing on bridging visual perception and symbolic planning.
• Developed a multimodal planning pipeline that transforms image observations and natural language task descriptions into PDDL representations for classical planners.
• Investigated scene graph generation, predicate extraction, and symbolic world modeling to improve the reliability of VLM-based planning.
• Reproduced and analyzed state-of-the-art methods including NL2Plan, Select2Plan, and formal visual planning frameworks, identifying limitations in spatial grounding and long-horizon reasoning.

Data Science in Data Science – University Of Information technology Ho Chi Minh (2023-08)

Deep Learning Specialization in Deep Learning – Coursera

IBM Data Science Professional Certificate in Data Science – Coursera

Top 40 leader board in Deep Learning – TensorTonic - Leetcode like for DL Models