Sulin Liu

Hi! I'm Sulin.

I am currently a post-training lead at Adobe foundation model team. Previously, I was a postdoc at MIT working with Rafael Gómez-Bombarelli and Tommi Jaakkola. I received my PhD from Princeton University, advised by Ryan P. Adams and Peter J. Ramadge. I went to National University of Singapore for my undergrad and worked with Sinno Jialin Pan at Nanyang Technological University.
My current interests are meta reward models, self-improving and multi-agent systems. Besides work and family, I like debugging my Gaggimate, studying Scaling Law for the Soul, sports hopping and testing AI succeed/fail.

Research

My work bridges theory and practice—I derive scalable learning objectives from first principles and implement them as robust, well-engineered systems.

Think While You Generate: Discrete Diffusion with Planned Denoising. Paper | Code | Video
Sulin Liu, Juno Nam, Andrew Campbell, Hannes Stärk, Yilun Xu, Tommi Jaakkola, Rafael Gómez-Bombarelli.
International Conference on Learning Representations (ICLR), 2025.

Flow matching for accelerated simulation of atomic transport in crystalline materials. Paper | Code
Juno Nam, Sulin Liu, Gavin Winter, KyuJung Jun, Soojung Yang & Rafael Gómez-Bombarelli.
Nature Machine Intelligence (2025).
Best paper, ICML Workshop on ML for Life and Material Science (2/141)

Generative Marginalization Models. Paper | Code | Video
Sulin Liu, Peter J. Ramadge, Ryan P. Adams.
International Conference on Machine Learning (ICML), 2024.
Contributed talk (6/125) at ICML Workshop on Structured Probabilistic Inference and Generative Modeling.

Sparse Bayesian Optimization. Paper | Code | Tutorial | Video
Sulin Liu* (equal contribution), Qing Feng*, David Eriksson*, Benjamin Letham, Eytan Bakshy.
International Conference on Artificial Intelligence and Statistics (AISTATS), 2023.
Contributed talk (top 5) at NeurIPS Workshop on Gaussian Processes, Spatiotemporal Modeling, and Decision-making Systems.

Task-Agnostic Amortized Inference of Gaussian Process Hyperparameters. Paper | Code | Slides | Video
Sulin Liu, Xingyuan Sun, Peter J. Ramadge, Ryan P. Adams.
Advances in Neural Information Processing Systems (NeurIPS), 2020.
Spotlight talk at 7th ICML Workshop on Automated Machine Learning.

View all publications →