Postdoctoral AI Researcher in Power Systems

Brookhaven Science Associates, LLC; Upton (NY); On-site; USD 70,200 - 85,000; Full time

Job summary

Brookhaven Science Associates, LLC invites applications for a one‑year term in the Energy and Photon Science Directorate to advance AI foundation models for electric applications focusing on GridFM. The role emphasizes developing scalable graph ML approaches, data generation, and rigorous benchmarking.

The selected candidate will work on GPU‑accelerated training, uncertainty quantification, and physics‑informed AI, with opportunities to extend the work through a renewal.

Qualifications

• Ph.D. required in a relevant field with demonstrated research capability.
• Strong background in machine learning and deep learning.
• Experience with PyTorch, JAX, TensorFlow, or similar frameworks.
• Some experience developing Graph Neural Networks (GNNs), Graph Transformers, or foundation‑model architectures.
• Familiarity with model training, fine‑tuning, evaluation, and deployment.
• Understanding of uncertainty quantification, model robustness, and physics‑informed AI.
• Experience with GPU computing and large‑scale model training.
• Demonstrated ability to conduct independent research.

Responsibilities

• Extend current GridFM capabilities for distribution networks.
• Develop scalable graph‑based machine learning or related models.
• Expand training data generation capabilities.
• Create benchmarks and test developed models.

Skills

Machine learning

Deep learning

PyTorch

JAX

TensorFlow

GNNs / Graph Transformers

Uncertainty quantification

Physics-informed AI

GPU computing

Independent research

Education

Ph.D. in Computer Science, Electrical Engineering, Mathematics, Physics, or a related field

Tools

Git

Docker

CI/CD

OpenDSS

GridLAB‑D

MATPOWER

PowerModels

PSS/E

Job description

Job Overview

The Energy and Photon Science Directorate advances basic science that underpins discoveries and breakthroughs for energy systems. The appointment is for a one-year term with an option for a one-year renewal, funded by project and performance. The successful candidate will contribute to the development of next‑generation AI foundation models and AI‑enabled workflows for electric applications, focusing on advancing GridFM, a grid foundation model for power systems.

Essential Duties and Responsibilities

• Extend current GridFM capabilities for distribution networks
• Develop scalable graph‑based machine learning or related models
• Expand training data generation capabilities
• Create benchmarks and test developed models

Required Knowledge, Skills, and Abilities

• Ph.D. in Computer Science, Electrical Engineering, Mathematics, Physics, or a related field.
• Strong background in machine learning and deep learning.
• Experience with PyTorch, JAX, TensorFlow, or similar frameworks.
• Some experience developing Graph Neural Networks (GNNs), Graph Transformers, or foundation‑model architectures.
• Familiarity with model training, fine‑tuning, evaluation, and deployment.
• Understanding of uncertainty quantification, model robustness, and physics‑informed AI.
• Experience with GPU computing and large‑scale model training.
• Demonstrated ability to conduct independent research.

Preferred Knowledge, Skills, and Abilities

• Familiarity with distributed computing, HPC environments, and cloud platforms.
• Experience building production‑quality software and ML pipelines.
• Familiarity with Git, CI/CD, containerization (Docker), and reproducible workflows.
• Experience developing APIs and workflow orchestration systems.
• Experience optimizing AI workloads for performance and scalability.
• Basic knowledge of electric power systems, transmission/distribution networks, power flow, optimal power flow, contingency analysis, or grid planning.
• Familiarity with tools such as PowerModels, MATPOWER, PSS/E, GridLAB‑D, OpenDSS, or similar.
• Experience with mathematical optimization, mixed‑integer programming, stochastic optimization, or decision analytics.
• Familiarity with Gurobi, CPLEX, Pyomo, JuMP, or related tools.
• Experience with LLM‑based workflows, tool‑calling agents, MCP architectures, retrieval systems, or AI copilots.
• Familiarity with multi‑agent systems and decision‑support applications.

Other Information

Candidates must have completed all degree requirements by the commencement of employment. BNL policy requires that after obtaining a Ph.D., eligible research associate appointments may not exceed a combined total of five years of relevant post‑doc and/or R&D experience, excluding time associated with family planning, military service, illness, or other life‑changing events.

The selected candidate must be able to obtain and maintain a DOE UPIV credential, as required by DOE Order 206.2 Chg. 2.

The base salary for this position ranges from $70,200 to $85,000 per year, commensurate with experience and peer group.