Postdoc: Scientific ML for Excited-State Chemistry

Lawrence Berkeley National Laboratory; Berkeley (CA); Hybrid; USD 99,000 - 114,000; Full time

Job summary

Berkeley Lab seeks a Postdoctoral Researcher to advance scientific machine learning and computational chemistry. You will build symmetry-aware models to accelerate excited-state simulations and connect them to open-source workflows.

Responsibilities include developing data-generation pipelines, validating model performance, and integrating software with nonadiabatic dynamics tools. PhD within 3 years and strong ML in science required.

Qualifications

• PhD in Computer Science, Computational Science, Chemistry, Physics or related field, within the last 3 years.
• Experience applying ML/deep learning to scientific or atomistic data.
• Strong Python programming and modern ML framework experience (PyTorch or JAX).

Responsibilities

• Conduct physics-informed, symmetry-aware ML for nonadiabatic excited-state dynamics.
• Develop equivariant graph networks learning potential-energy surfaces and forces.
• Design data-generation and training workflows with active/transfer learning and uncertainty analysis.
• Implement and maintain research software; integrate models with nonadiabatic workflows.
• Plan ablation studies with reproducible evaluations across seeds and metrics.
• Collaborate to connect models to established workflows like DeepMD, NEXMD, i-PI.
• Publish results and mentor students/researchers.

Skills

Python programming

PyTorch

JAX

Graph neural networks

Linux environment

Version control

Reproducible workflows

Education

PhD degree

Tools

PyTorch

JAX

Linux

Job description

Berkeley Lab seeks a Postdoctoral Researcher to advance scientific machine learning and computational chemistry. You will build symmetry-aware models to accelerate excited-state simulations and connect them to open-source workflows.

Responsibilities include developing data-generation pipelines, validating model performance, and integrating software with nonadiabatic dynamics tools. PhD within 3 years and strong ML in science required.