Jinwoo Go

Brookhaven National Laboratory

Upton, NY 11973

I am a Research Staff in the Computing and Data Sciences Division at Brookhaven National Laboratory. My research focuses on uncertainty-aware scientific discovery using machine learning, spanning Bayesian optimal experimental design, generative modeling, and decision-making under uncertainty. My long-term goal is to build AI systems that actively guide data acquisition, reason under uncertainty, and support robust decisions in complex scientific and engineering problems.

My work spans three interconnected directions:

Goal-driven Bayesian experimental design. I develop frameworks that optimize experiments for downstream scientific decisions rather than parameter estimation alone, with applications in risk-aware epidemiological control and intervention planning.
Generative models for scientific inference and design. I develop diffusion- and flow-matching-based methods for efficient posterior sampling and uncertainty quantification, with applications to inverse problems and optimal control in scientific domains, including power-grid systems.
LLM-informed scientific priors. I study how large language models and foundation models can serve as structured, domain-aware priors to improve Bayesian inference and experimental design.

At Brookhaven, I collaborate with Dr. Byung-Jun Yoon, Dr. Xiaoning Qian, and students on machine learning methods for scientific computing and decision-making under uncertainty.

I received my Ph.D. in Computational Science and Engineering from the Georgia Institute of Technology. My doctoral work focused on scalable Bayesian inference, inverse problems, and optimal experimental design for computational science.

selected publications

JCP

Sequential Infinite-Dimensional Bayesian Optimal Experimental Design with Derivative-Informed Latent Attention Neural Operator

Jinwoo Go and Peng Chen

Journal of Computational Physics, 2025

Abs HTML

We propose a sequential Bayesian optimal experimental design framework for infinite-dimensional parameter spaces using a latent-variable attention-based neural operator (LANO) surrogate for time-dependent PDEs, enabling scalable and efficient design optimization.
CMAME

Accurate, Scalable, and Efficient Bayesian Optimal Experimental Design with Derivative-Informed Neural Operators

Jinwoo Go and Peng Chen

Computer Methods in Applied Mechanics and Engineering, 2025

Abs HTML

We improve the computational efficiency of Bayesian optimal experimental design using derivative-informed neural operators (DINO), achieving accurate and scalable design optimization for high-dimensional problems.
UAI

Robust Expected Information Gain for Optimal Bayesian Experimental Design Using Ambiguity Sets

Jinwoo Go and Tobin Isaac

In Proceedings of the 38th Conference on Uncertainty in Artificial Intelligence, 2022

Oral Presentation — Top 5%

Abs HTML

We propose a distributionally robust formulation of expected information gain for Bayesian experimental design using ambiguity sets, enabling risk-averse experiment selection under prior uncertainty.