PhD scholarships and research opportunities.
The scholarship page is the main place for current opportunities across AI Future, AI for Science, and AI for Health, including what to send and how to make contact.
View scholarships
Professor at Deakin University
Building AI systems that reason, code, plan, and support professional work in the real world.
I am Truyen Tran, Professor and Head of AI, Health and Science at the Applied Artificial Intelligence Institute, Deakin University. My work spans foundational AI, AI for scientific discovery, and AI for health, with a growing focus on the parts of intelligence that remain distinctly human: agency, critical thinking, judgement, and taste.
My group develops AI systems that combine language, perception, retrieval, tools, coding, and evaluation, then asks how those systems should be directed, critiqued, and governed when the hardest layer is no longer raw capability but good judgement.
Current focus
From models to systems: architectures that retrieve, reason, code, plan, propose, evaluate, and interact with domain experts in the loop.
Students and Visitors
PhD opportunities and the reading schedule.
Key information for prospective students, collaborators, and anyone following the reading group.
Current session timing and upcoming topics.
The reading page is maintained as a practical schedule page so people can quickly check the exact time, location, upcoming talks, and recent reading club activity.
View reading schedule
AI Era
AI has entered a phase transition. Research and training must change with it.
Coding, implementation, and many early-stage research moves are now AI-native. The point is no longer to compete with the tool on routine work, but to work one level above it.
Read the full manifestoTreat AI as baseline
AI-generated ideas and implementations are treated as the new baseline, not as surprising exceptions. Real novelty starts one step beyond what the best available AI can already propose.
Redefine the PhD
A PhD is still about novelty and substance, but now it also requires independent investigation in an AI-rich environment: conceiving problems, directing agents, judging results, and building work with lasting value.
Future-proof skills
The skills with the longest shelf-life are agency, critical thinking, judgement, taste, mathematics, philosophy, product sense, and the ability to invent new directions rather than chase incremental gains.
Active Systems
Active systems and research infrastructure taking shape here.
A working environment for AI-native discovery, invention, proof, education, and deployment.
A multi-expert AI operating environment
A growing ecosystem of specialist agents for strategy, invention, policy, scientific reasoning, and domain advice, designed to coordinate serious work rather than produce one-off demos.
Discovery and theorem engines
Scientific discovery workflows and AI-assisted mathematics, including hypothesis generation, rediscovery benchmarks, theorem handoff, and formal verification through Lean.
Efficient and embodied intelligence
Work on energy-aware AI, evidence-centric clinical generation, and multi-sense architectures that connect perception, memory, planning, action, and safety.
Virtual PhD and professor tooling
Research-management systems that treat AI as part of the research team: literature analysis, ideation, experiment planning, supervision, reporting, and strategic judgement.
GrantOS and research strategy systems
Code-first grant and research tooling for opportunity diagnosis, proposal shaping, reviewer simulation, and long-horizon program design.
Research Platforms
Three flagship programs, one integrated research identity.
The portfolio is designed to connect frontier method development with systems that matter in laboratories, clinics, and public life.
Foundations for next-generation intelligence
Capabilities are no longer the main story on simple tasks. The frontier is agency, critical thinking, judgement, taste, and trustworthy action in open-ended environments, alongside long-horizon work on the physics of intelligence, low-energy AI, full embodiment, and post-Transformer alternatives.
Reliable agents
Physics of intelligence
Memory
Embodiment
Quantum systems
Explore platform
Scientific copilots for automated discovery
Systems for materials, molecules, mathematics, and experimental work where the bottleneck is no longer text generation, but scientific rigor, iteration speed, and integration with real research workflows.
Materials
Drug design
Scientific reasoning
Explore platform
Clinically useful AI systems
Clinical AI systems for imaging, longitudinal health, monitoring, and decision support where the hard problems are deployment, trust, workflow fit, and measurable patient benefit.
Clinical AI
Imaging
Longevity
Explore platform
Systems We Build
Agentic and decision-support systems with technical depth.
The challenge is no longer whether models can answer or code in isolation, but whether systems can exercise agency, support critical thinking, and defer to human judgement in real work.
Research planning, experiment selection, code-assisted analysis, evaluation, and scientific communication in large discovery spaces.
Clinical support systems that observe, retrieve, reason, and support decisions with safety and human oversight in mind.
Temporal and multimodal representations of patient state for monitoring, intervention timing, and long-range care planning.
Agentic workflows for research synthesis, coding, experimentation, evaluation, and collaboration across teams and domains.
Current Focus
Research directions shaping the next phase.
A snapshot of active themes across the group, spanning foundational architectures and high-consequence applications.
Agentic AI for scientific discovery
Systems that can reason scientifically, write code, coordinate tools, and accelerate discovery loops.
Clinical world models
Representations of longitudinal patient state built for intervention timing, monitoring, and decision support under real clinical constraints.
Agency and judgement architectures
Architectures that combine model inference with retrieval, memory, planning, and explicit evaluation loops while keeping human judgement in the loop.
Materials and molecular design
Generative, retrieval-based, and simulation-linked systems for materials, molecules, and inverse design where validation cost is high.
Judgement and taste in deployed AI
Alignment, oversight, value sensitivity, decision quality, and practical governance for deployed AI.
AI for global health and biomedicine
Scalable tools for imaging, mental health, public health, and resource-constrained settings.
Selected Impact
Proof points across research, funding, and public engagement.
Selected signals across research, funding, and public engagement provide a concise view of the work and its reach.
AI Future, AI for Science, and AI for Health define the portfolio.
Including MRFF-supported programs in cardiovascular health and AI in mental health.
Talks and keynotes include Los Alamos, RIVF, VNU Hanoi, and public lecture circuits.
Work With Me
Clear pathways for students, collaborators, partners, and organisers.
I work with students, researchers, clinicians, industry, government, and global partners who want to build high-impact AI systems with real-world consequence.
Join as student
For prospective PhD students and postdocs interested in reliable AI systems, scientific discovery, health AI, and real deployment constraints.
What to sendCollaborate on research
For academics and labs exploring shared questions in AI systems, science, medicine, and translational methods.
Research collaborationsPartner on impact
For industry, health, government, and defence partners seeking technically serious collaboration pathways.
Partnership routes
Recent Writing
Thought leadership across AI, science, health, and society.
- AI, math, medicine, software, and the sciences
A shifting landscape for interdisciplinary AI practice. - A.I development and education
Interview and reflections on education, capability, and long-term consequences. - Vietnamese writing and public-facing essays
Curated writing across English and Vietnamese channels.
Recent Talks
Curated talks rather than a raw chronological archive.
- AI for scientific discovery: A new frontier
RIVF keynote, December 2025. - AI4Science: The 5th paradigm
Los Alamos National Laboratory, September 2025. - AI: The tipping point
VASEA webinar, March 2025.