Postdoctoral Scholar – AI for Functional Polymer Discovery

Lawrence Berkeley National Laboratory • Berkeley, CA

10d • Onsite

About The Position

Lawrence Berkeley National Laboratory is hiring a Postdoctoral Scholar – AI for Functional Polymer Discovery within the Molecular Foundry division to support a newly funded two-year project focused on AI-driven discovery of high-performing polymer dielectrics for next-generation power electronics.
In this role, the Postdoctoral Researcher will develop and implement a multimodal AI–digital twin framework that connects generative polymer design with physics-based multiscale simulations, machine-learning surrogate models, and experimental validation in a closed-loop workflow. The work will integrate machine learning, quantum chemistry, molecular dynamics, and experimental feedback, leveraging Berkeley Lab’s computational resources and the Foundry’s capabilities in polymer synthesis, processing, and dielectric characterization, within a highly interdisciplinary environment with opportunities for high-impact publications, open datasets, and prototype demonstrations.

Requirements

• Ph.D. (within the last two years) in Materials Science, Polymer Science, Chemistry, Chemical Engineering, Physics, or a related field.
• Strong background in at least one of the following areas: Molecular dynamics simulation and quantum chemistry calculation on polymers
• Machine learning applied to materials or chemistry (e.g., GNNs, Generative Models).
• Dielectric materials or functional polymers
• Demonstrated ability to conduct independent research and collaborate in interdisciplinary teams.
• Strong written and verbal communication skills, evidenced by peer-reviewed publications.

Nice To Haves

• Experience with machine learning frameworks (e.g., PyTorch, TensorFlow) and transformer-based architectures;
• Familiarity with high-throughput molecular dynamics, DFT, or ML-based interatomic potentials (e.g., DeepMD, MACE);
• Experience working with polymer synthesis, processing, or dielectric characterization;
• Experience with active learning, uncertainty quantification (UQ), multimodal data fusion, model interpretability methods (e.g., SHAP);
• Experience working in collaborative environments.
• -Demonstrated research software engineering practices (clean code, Git, testing, packaging/workflow automation).
• Practical HPC experience (schedulers, scaling runs, workflow tools like Snakemake/Parsl/FireWorks—any similar evidence).
• Familiarity with FAIR-ish data practices (metadata standards, reproducibility, dataset governance).
• Experience collaborating with experimentalists and translating computational results into experimental decisions.

Responsibilities

• Develop and implement generative molecular design workflows for polymer dielectrics using reaction-aware chemical rules, monomer libraries, and transformer-based chemical language models.
• Perform physics-based simulations (e.g., molecular dynamics, DFT) and implement Machine-Learned Interatomic Potentials (MLIPs) to predict physical parameters.
• Build and train machine-learning surrogate models (e.g., GNNs).
• Integrate simulation, ML, and experimental results into a closed-loop AI-digital twin framework utilizing uncertainty-guided active learning to improve predictive accuracy.
• Collaborate with experimental scientists to translate model predictions into synthesis and testing priorities.
• Analyze and interpret structure–property relationships using interpretable ML tools such as SHAP or attention maps.
• Communicate research progress through internal presentations, external conference talks, and peer-reviewed publications.
• Contribute to open datasets, codes, and best practices supporting reproducible AI-enabled materials discovery.

What This Job Offers

• Job Type: Full-time
• Career Level: Entry Level
• Education Level: Ph.D. or professional degree
• Number of Employees: 101-250 employees