Hiring! I am looking for self-motivated PhD/Master students to join my group at SMU to work on i) fundamental decision making (i.e., single- and multi-agent reinforcement learning), ii) decision making for applications (e.g., RL for FinTech, MARL for urban security, AI for science), and iii) foundation agents (e.g., computer control agents and auto research agents). The requirement for SMU PhD admission can be found here. Visiting CSC students/researchers and remote interns are also welcome — please email me if you are interested.
About Me
I am an Assistant Professor and Lee Kong Chian Fellow at the School of Computing and Information Systems, Singapore Management University, where I lead the DIGA research group. My work builds highly-capable agents — empowered by (multi-agent) reinforcement learning and (multimodal) large language models — for complex decision-making tasks, and applies them to finance, urban security, and science.
Research Interests
(Multi-Agent) (Reinforcement) Learning, (Distributed) Foundation Models/Agents, Computer Science of Language Models, and AI for Science.
Pages we actively maintain:
MatVerse Paper Collection
A regularly updated paper collection on AI for science, particularly for materials science.
Computer Science of Language Models
Language models are becoming the new universal computer, solving even more problems than traditional computers — yet the computer science for them is still missing.
Agent Cybernetics
Building generalist agents by drawing inspiration from cybernetics — a transdisciplinary approach to regulatory systems, their structures, constraints, and possibilities.
News
- Apr 2026 FinMaster is accepted to ACL 2026 (Findings). Congrats to Junzhe and Chang!
- Mar 2026 LLM-Based World Models Can Make Decisions Solely, But Rigorous Evaluations are Needed is accepted to TMLR. Congrats to Chang!
- Jan 2026 NPPC is accepted to TMLR. Congrats to Chang!
- Nov 2025 We are awarded USD 50,000 in Google Cloud credits through the Gemini Academic Program to support our research using Google's Gemini models!
- Nov 2025 We release the paper MATAI: A Generalist Machine Learning Framework for Property Prediction and Inverse Design of Advanced Alloys. Comments are welcome!
- Sep 2025 We release a paper collection on AI for (materials) science to facilitate the research. It is updated regularly — please contact me for potential collaborations.
- Jul 2025 We release the paper AutoMAT: A Hierarchical Framework for Autonomous Alloy Discovery, our first paper on AI for science. Comments are welcome!
- May 2025 We release the paper FinMaster: A Holistic Benchmark for Mastering Full-Pipeline Financial Workflows with LLMs. Comments are welcome!
- Apr 2025 We release the paper Nondeterministic Polynomial-time Problem Challenge: An Ever-Scaling Reasoning Benchmark for LLMs. Comments are welcome!
- Nov 2023 I will join Singapore Management University (SMU) as an Assistant Professor in July, 2024.
- Mar 2023 We release our reinforcement learning for FinTech platform TradeMaster.
AI for Materials Science
MATAI Platform
An interactive platform for AI-driven alloy discovery.
AutoMAT
Autonomous multi-objective alloy design through simulation-guided optimization.
MATAI
A generalist machine learning framework for property prediction and inverse design of advanced alloys.
AutoDFT
A closed-loop multi-agent framework for autonomous DFT calculations.
MatSeek
An automated knowledge-driven framework for materials research.
Speculations
- All decision-making scenarios — single-agent, cooperative multi-agent, competitive multi-agent (i.e., game theory), and mixed cooperative-competitive — will be mastered through one algorithm, one model, and even one set of parameters of the model.
- All the methods improving over a base foundation model — i.e., prompt engineering such as CoT, self-reflection, and alignment — should be addressed by a universal method, maybe reinforcement learning (RL).
- For any given problem A (maybe NP-hard), we can find another NP problem B such that if the foundation model can solve B, it can solve A. Therefore, AGI/ASI can be achieved by focusing on NP problems only.
- The foundation agent should emerge from a foundation model capable of solving difficult tasks. Structurally, it comprises three components: (i) a highly capable multimodal language model, (ii) minimal embodiment for tool creation and use, and (iii) optimization via reinforcement learning to drive the emergence of reasoning and tool-related capabilities. Mechanistically, it follows a trinity architecture: (i) a multimodal LLM, (ii) a memory module, and (iii) a world model, where memory and world model together form a universal harness that grounds and contextualizes the agent's behavior.
- Artificial General Intelligence (AGI) will ultimately be i) self-improving, i.e., searching with regular retraining, on decisions and alignment through interacting with (virtually) embodied environments such as video games, and ii) distributed intelligently across massive open-sourced domain-specific models.
- The current AI revolution represented by LLMs is nearing its end in the service industry, or the tertiary sector. Only by entering the secondary sector, represented by industry and manufacturing, can AGI ultimately be achieved. This will be piloted by AI for science, with AI for manufacturing as the main force.