The research group of Prof. Vincent Y. F. Tan (https://vyftan.github.io/) at NUS invites applications for multiple postdoctoral research fellow positions:
• Multi-Armed Bandits and Online Decision Making
• Reinforcement Learning (RL)
• Information Theory for Machine Learning and Language Models
• Statistical Signal Processing
These positions are supported in part by the prestigious National Research Foundation (NRF) Investigatorship (2026) awarded for the project:
“Fundamental Limits of Language Models via Information Theory.
Research Vision
Our group aims to develop a mathematically rigorous foundation for modern learning systems, particularly large language models (LLMs), by combining:
- Information-theoretic limits
- Sequential decision-making theory
- Optimization
Recent directions include:
- Provable limits of transformers and prompt tuning
- Sample complexity and sharp minimax bounds for heteroscedastic and structured bandits
- Offline RLHF and preference learning
- Parameter-free online optimization algorithms
- Adaptive decoding and bandit-based inference for LLMs
Our work appears regularly in leading venues such as International Conference on Machine Learning (ICML), International Conference on Learning Representations (ICLR), Neural Information Processing Systems (NeurIPS), and ÌÇÐÄlogo Transactions on Information Theory.
We are particularly interested in postdocs who are excited about bridging information theory and modern AI, and who enjoy proving sharp theorems while staying close to real machine learning systems.
Environment
The positions are jointly hosted by the Department of Mathematics, National University of Singapore. NUS is consistently ranked among the top universities globally and is the leading university in Asia. Singapore offers a vibrant international research environment and strong connections to industry research labs.
Candidate Profile
We seek candidates who:
- Have (or are about to obtain) a Ph.D. in Electrical Engineering, Computer Science, Mathematics, Statistics, or related fields
- Possess a strong publication record in top-tier venues
- Have expertise in at least one of:
- Reinforcement learning or bandits
- Information theory
- Online convex optimization
- Learning theory
- Large language models with theoretical analysis
- Are mathematically mature or with a strong programming background.
Strong candidates with independent research vision will be encouraged to develop their own research directions within the broader theme of the grant.
What We Offer
- Competitive salary (commensurate with experience)
- Travel support to top conferences
- Mentorship toward academic or research-lab careers
- Opportunity to work on a flagship NRF Investigatorship project
- Collaboration opportunities with leading international researchers
Former postdocs and students from the group have gone on to prestigious faculty and research positions worldwide.
How to Apply
Interested candidates should email:
with:
- CV
- Research statement
- Selected publications
- Names of referees
Inquiries are welcome. Review of applications will begin immediately and continue until positions are filled.
If you are excited about establishing fundamental limits for learning systems and language models—and about proving theorems that shape the future of AI—we encourage you to apply.