Advancing the frontier of trustworthy AI
We research, build, and openly share the next generation of intelligent systems.
Why we exist
Our mission is to advance the science of machine intelligence and translate it into open tools, datasets, and benchmarks that the whole community can build on.
Where we're going
A future where AI systems are capable, transparent, and trustworthy — developed in the open and accountable to the people they serve.
Research directions
Language Models & Reasoning
Building and evaluating large language models that reason reliably and transparently.
Computer Vision
Robust visual understanding that holds up under real-world distribution shift.
Trustworthy & Open AI
Open datasets, benchmarks, and methods for safe, accountable AI systems.
Highlights & latest news
DreamAI has 12 papers accepted at EMNLP 2026 conference (A*)
DreamAI has 12 papers accepted at EMNLP 2026 conference (A*)
Read more →Paper accepted at NeurIPS 2025
Our work on trustworthy reasoning in LLMs was accepted as an oral presentation.
Read more →Releasing the RobustVQA dataset
A new open dataset for evaluating robust visual question answering.
Read more →Core team
Trung Le
Senior Lecturer
AI/ML · Human-AI Alignment · Machine Reasoning · ...
Dr. Trung Le is a Senior Lecturer (U.S. Associate Professor equivalent) at Monash University and a co-founder of the DreamAI Research Lab. My work bridges foundational theories, such as optimal transport, optimization, Bayesian and probabilistic inference, information theory, and causality, with next‑generation AI challenges including LLM alignment, reasoning, continual learning, generative AI, and trustworthy machine learning. My long‑term vision is to develop smart, intelligent, and secure AGI foundation models that are equitably accessible to everyone, and to mentor and cultivate the next generation of bold, rigorous AI researchers in Australia and Vietnam.
Join DreamAI
We welcome students, researchers, and collaborators eager to build trustworthy, open AI. Get in touch to work with us.