Postdoctoral Appointee - Artificial Intelligence Modeling - Hybrid

About Sandia:

Sandia National Laboratories is the nation’s premier science and engineering lab for national security and technology innovation, with teams of specialists focused on cutting‑edge work in a broad array of areas. Some of the main reasons we love our jobs:

• Challenging work with amazing impact that contributes to security, peace, and freedom worldwide
• Extraordinary co‑workers
• Some of the best tools, equipment, and research facilities in the world
• Career advancement and enrichment opportunities
• Flexible work arrangements for many positions include 9/80 (work 80 hours every two weeks, with every other Friday off) and 4/10 (work 4 ten‑hour days each week) compressed workweeks, part‑time work, and telecommuting (a mix of onsite work and working from home)
• Generous vacation, strong medical and other benefits, competitive 401k, learning opportunities, relocation assistance and amenities aimed at creating a solid work/life balance*

World-changing technologies. Life-changing careers. Learn more about Sandia at: http://(Use the "Apply for this Job" box below)..gov

* These benefits vary by job classification.

Sandia’s artificial intelligence (AI) team is building the U.S. Department of Energy’s (DOE) next‑generation AI Platform, an integrated scientific AI capability that delivers rapid, high‑impact solutions for national security, science, and applied energy missions. The Platform is based on three pillars:
Models, Infrastructure, and Data. As a Postdoctoral Appointee, you will join the Models Pillar team to architect, develop, and deploy fine‑tuned reasoning models, domain foundation models, high‑fidelity surrogate models, and autonomous agents. Your work will compress mission timelines by enabling scientists and engineers to explore design spaces, evaluate outcomes, and steer experiments and simulations with transparent, high‑assurance AI workflows.

We anticipate multiple hires for the Models Pillar that collectively span the set of responsibilities and skills described below. Likewise, postdoctoral appointees will be expected to work in conjunction with their Sandia mentors and teams from across Sandia and other DOE laboratories to deliver on this ambitious, fast‑paced project. Importantly, we anticipate that while AI Platform development will leverage existing AI and data science tools extensively, success will also require deep technical insights, considerable innovation, research, and problem solving to address the unique needs of DOE applications.

If this sounds like an exciting challenge to you, we look forward to reading your application!

• Research, fine‑tune, and certify large reasoning models (LLMs, graph neural nets, vision transformers, etc.) for domain tasks in materials science, chemistry, physics, grid controls, and nuclear security
• Develop and integrate domain foundation models trained or adapted on DOE simulation, experimental, and production data
• Build AI surrogates to accelerate exascale multiphysics simulations, enabling millisecond‑scale predictions
• Design and implement multi‑agent frameworks (hypothesizers, planners, executors, retrievers, assessors) with transparent decision graphs, uncertainty quantification, and audit logs
• Embed continuous learning pipelines: connect model training/evaluation to live telemetry from HPC clusters, experiments, and autonomous labs
• Develop a model repository with metadata, SBOMs, versioning, drift/poisoning surveillance, and periodic recertification
• Develop and implement high‑assurance controls: least‑privilege execution, runtime shields/tripwires, deterministic fallbacks, cryptographic provenance, and enclave attestation for sensitive workloads
• Collaborate with Data and Infrastructure teams to align model requirements with data lake houses, compute fabric, and edge inference systems
• Contribute to open‑source and internal AI frameworks, toolkits, and best practices for agentic workflows

• Prototype a custom transformer for multisensor fusion in an agile‑deterrence scenario
• Optimize a surrogate neural network to replace a costly physics submodule in a reactor design simulation
• Design a Planner agent…