Curriculum Vitae
Download PDF
Ph.D. Candidate in Computer Science, National University of Singapore · Jan 2023 – Jan 2027
yannis.montreuil@u.nus.edu ·
Google Scholar ·
GitHub ·
LinkedIn
Research Profile
Final-year PhD student at NUS with seven first or co-first author papers at NeurIPS, ICLR, ICML, and AISTATS. I work on decision-making under uncertainty: learning, for each input, which model to trust, when to combine several, and when none is reliable. I have applied this to time series with regime changes, LLMs, online learning, and adversarial robustness, with proofs and out-of-sample experiments in Python/PyTorch.
I expect to complete my PhD in January 2027. I am looking for full-time roles and internships in quantitative research or machine learning research.
Finance & Markets
- Jump Trading AI Summit: selected participant, Hong Kong, Oct 2026.
- Optiver PhD Quant Focus: selected participant, quantitative research track, Sydney, Jul 2026.
- Crypto trading: started in 2016 on early exchanges (Kraken, Bittrex).
Publications
Peer-reviewed
- Beyond Augmented-Action Surrogates for Multi-Expert Learning-to-Defer. Yannis Montreuil, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. NeurIPS 2026. Link
- Why Ask One When You Can Ask k? Learning-to-Defer to the Top-k Experts. Yannis Montreuil, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. ICLR 2026. Link
- Online Learning-to-Defer with Varying Experts. Yannis Montreuil*, Dang Hoang Duy*, Maxime Meyer*, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. AISTATS 2026. Link
- Adversarial Robustness in One-Stage Learning-to-Defer. Yannis Montreuil*, Letian Yu*, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. AISTATS 2026. Link
- Optimal Query Allocation in Extractive QA with LLMs: A Learning-to-Defer Framework with Theoretical Guarantees. Yannis Montreuil*, Yeo Shu Heng*, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. AISTATS 2026. Link
- Adversarial Robustness in Two-Stage Learning-to-Defer: Algorithms and Guarantees. Yannis Montreuil, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. ICML 2025. Link
- A Two-Stage Learning-to-Defer Approach for Multi-Task Learning. Yannis Montreuil*, Yeo Shu Heng*, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. ICML 2025. Link
- Towards Robust Human–AI Decision-Making via Learning-to-Defer. Yannis Montreuil. AAAI-26 Doctoral Consortium.
Preprints
- A Query Is Not a Commitment: Learning to Correct Expert Answers in Online Deferral. Yannis Montreuil, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. Preprint. Link
- Consistent Learning-to-Defer with Expert-Conditional Advice. Yannis Montreuil, Leina Montreuil, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. ICML 2026 AgenticUG Workshop; under review. Link
- Learning to Defer in Non-Stationary Time Series via Switching State-Space Models. Yannis Montreuil, Letian Yu, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi. Under review. Link
- Transition Affinity: A Unified Notion of Geometry for LLMs and Text Corpora. Maxime Meyer*, Yannis Montreuil*, Zongmin Li*, Caroline Chaux, Vincent Y. F. Tan. Under review.
* indicates equal contribution.
Research Experience
Doctoral Researcher, Computer Science
- Designed surrogate losses with consistency guarantees for selecting and combining the top-k experts per query (NeurIPS 2026, ICLR 2026, ICML 2025).
- Built switching state-space models that detect regime changes in non-stationary time series and adapt which predictor to trust (under review).
- Developed online routing algorithms that keep learning as models are added, removed, or change in quality (AISTATS 2026).
- Built cost-aware routers that send each query to the most suitable LLM or expert, balancing answer quality and inference cost (AISTATS 2026, ICLR 2026).
- Designed attacks on multi-model systems and adversarial training methods that provably resist them (ICML 2025, AISTATS 2026).
- Implemented all methods in Python/PyTorch, including data pipelines, multi-GPU training, and benchmark experiments.
Research Associate
- Contributed to a model-routing API for the DesCartes program (CNRS@CREATE).
- Collaborated with Prof. Wei Tsang Ooi, Prof. Axel Carlier, and Dr. Lai Xing Ng.
Research Intern
- Trained self-supervised deep models on 3D point clouds for robotic manipulation of deformable objects.
Research Intern
- Raised building-detection accuracy on satellite images by 30% with semi-supervised learning (Mean Teacher, Naive Student).
Education
Ph.D. in Computer Science
A*STAR Merit Award; AISG–DesCartes Joint Excellence Scholarship. Coursework: Machine Learning Theory; Probabilistic Graphical Models; Theoretical Computer Science; Trust, Privacy and Robustness.
M.Sc. in Industrial Engineering and Operations Research, Highest Honors
M.Sc. in Robotics, Honors
B.Sc. in Mechanical Engineering
Awards
Academic Service
Reviewer: NeurIPS 2026, ICML 2026 (Silver Reviewer), ICLR 2026, AISTATS 2026, TACL 2026.
Teaching: CS5242 Neural Networks and Deep Learning; CP2107 Independent Introduction to CS Research.
Students supervised: Yeo Shu Heng, Dang Hoang Duy, Letian Yu, Seah Min-Rei.
Technical Skills
| Mathematics & statistics | Probability, statistical inference, time series, state-space models, online learning, linear algebra, convex optimization. |
|---|---|
| Machine learning | Deep learning, robustness under distribution shift, out-of-sample evaluation, LLMs (Hugging Face Transformers). |
| Engineering | Python, PyTorch, NumPy, SciPy, multi-GPU training, Linux, Docker, GCP, Git; C, R; MOSEK. |
| Languages | French (native), English (fluent), Spanish (intermediate). |
| Interests | Muay Thai, tennis, table tennis, League of Legends when I find the time. |