Academic Work

RESEARCH.

Using machine learning and AI to automate and accelerate the design of chemical processes — from molecules to flowsheets.

Machine Learning· Graph Neural Networks· Reinforcement Learning· Chemical Engineering· Process Design· Multi-Agent Systems· Molecular Property Prediction· TU Delft· Machine Learning· Graph Neural Networks· Reinforcement Learning· Chemical Engineering· Process Design· Multi-Agent Systems· Molecular Property Prediction· TU Delft·
Focus Areas

RESEARCH
INTERESTS.

01

Machine Learning &
Deep Learning

Developing graph neural networks and self-supervised learning methods to predict molecular properties, environmental impacts, and polymer behavior directly from molecular structures — bypassing expensive experiments.

02

Reinforcement
Learning

Using reinforcement learning and transfer learning to automate chemical process design — from generating flowsheets to optimizing synthesis across simulation fidelities. Teaching AI to think like an engineer.

03

Process Design &
Chemical Engineering

Bridging AI and chemical engineering to tackle real-world challenges: sustainable process synthesis, life cycle assessment, and multi-agent systems for complex engineering workflows.

Selected Work

PUBLICATIONS.

Selected publications. For the full list, see Google Scholar.

2026
Multi-agent systems for chemical engineering: A review and perspective
S. Rupprecht, Q. Gao, T. Karia, A.M. Schweidtmann
Current Opinion in Chemical Engineering, Vol. 51, 101209, 2026
Reviews how teams of AI agents can collaborate to tackle complex chemical engineering tasks, and discusses future opportunities.
Paper →
2025
Accelerating process synthesis with reinforcement learning: Transfer learning from multi-fidelity simulations and variational autoencoders
Q. Gao, H. Yang, M.F. Theisen, A.M. Schweidtmann
Computers & Chemical Engineering, Vol. 201, 109192, 2025
Speeds up chemical process design by combining reinforcement learning with transfer learning across simulations of different detail levels.
Paper →
2025
Environmental impacts prediction using graph neural networks on molecular graphs
Q. Gao, L.S. Balhorn, A. Laera, R. Meys, J. Goßen, J.M. Weber, G. Wernet, A.M. Schweidtmann
Computers & Chemical Engineering, 109362, 2025
Predicts how chemicals affect the environment by applying graph neural networks directly to molecular structures.
Paper →
2025
Bayesian uncertainty quantification of graph neural networks using stochastic gradient Hamiltonian Monte Carlo
Q. Gao, D.C. Miedema, Y. Zhao, J.M. Weber, Q. Tao, A.M. Schweidtmann
Systems and Control Transactions, 1360–1364, 2025
Adds reliable uncertainty estimates to graph neural network predictions using a Bayesian sampling method.
Paper →
2024
Deep reinforcement learning for process design: Review and perspective
Q. Gao, A.M. Schweidtmann
Current Opinion in Chemical Engineering, Vol. 44, 101012, 2024
Reviews the state of deep reinforcement learning for chemical process design and outlines key future research directions.
Paper →
2024
Self-supervised graph neural networks for polymer property prediction
Q. Gao, T. Dukker, A.M. Schweidtmann, J.M. Weber
Molecular Systems Design & Engineering, Vol. 9(11), 1130–1143, 2024
Uses self-supervised learning on graph neural networks to predict polymer properties without needing large amounts of labeled data.
Paper →
2023
Flowsheet generation through hierarchical reinforcement learning and graph neural networks
L. Stops, R. Leenhouts, Q. Gao, A.M. Schweidtmann
AIChE Journal, Vol. 69(1), e17938, 2023
Automatically builds chemical process flowsheets by combining hierarchical reinforcement learning with graph neural networks.
Paper →
2023
Transfer learning for process design with reinforcement learning
Q. Gao, H. Yang, S.M. Shanbhag, A.M. Schweidtmann
Computer Aided Chemical Engineering, Vol. 52, 2005–2010, 2023
Shows that transferring knowledge between tasks significantly speeds up reinforcement learning for chemical process design.
Paper →
2021
Modeling category-selective cortical regions with topographic variational autoencoders
T.A. Keller, Q. Gao, M. Welling
arXiv preprint arXiv:2110.13911, 2021
Models how the brain organizes visual information into specialized regions using topographic variational autoencoders.
ArXiv →
Background

THE
JOURNEY.

2021–
Present
PhD Researcher — Process Intelligence
Delft University of Technology, Netherlands
Research focus: machine learning for molecular property prediction, reinforcement learning for chemical process design, and multi-agent systems.
2019–
2021
M.Sc. Computational Science
University of Amsterdam, Netherlands
Graduated with distinction. Thesis on topographic variational autoencoders for modeling cortical organization.
2015–
2019
B.Sc. Applied Chemistry
Northeast Forestry University, China
Foundation in polymer science, chemistry, and computer science.