Sandia National Laboratories · 13 hours ago
Postdoctoral Appointee - Artificial Intelligence Data Science - Hybrid
Albuquerque, NM
Full-time
Hybrid
New Grad, Entry LevelSandia National Laboratories is the nation’s premier science and engineering lab for national security and technology innovation. The Postdoctoral Appointee will join the AI team to design and operate an AI-ready data ecosystem, transforming various data types into governed datasets that support AI models and workflows.
National DefenseGovernmentInformation TechnologyNational Security
No H1B
Security Clearance RequiredU.S. Citizen Only
Responsibilities
Build and operate an AI-Ready Lakehouse
Design and maintain a federated data lakehouse with full provenance/versioning, attribute-based access control, license/consent automation, and agent telemetry services
Implement automated, AI-mediated ingestion pipelines for heterogeneous sources (HPC simulation outputs, experimental instruments, robotics, sensor streams, satellite imagery, production logs)
Enforce Data Security & Assurance
Develop a Data Health & Threat program: dataset fingerprinting, watermarking, poisoning/anomaly detection, red-team sampling, and reproducible training manifests
Configure secure enclaves and egress processes for CUI, Restricted Data, and other sensitive corpora with attestation and differential-privacy where required
Define and Implement Data Governance
Establish FAIR-compliant metadata standards, data catalogs, and controlled-vocabulary ontologies
Automate lineage tracking, quality checks, schema validation, and leak controls at record-level granularity
Instrument AI Workflows with Standardized Telemetry
Deploy Agent Trace Schema (ATS) and Agent Run Record (ARR) frameworks to log tool calls, decision graphs, human hand-offs, and environment observations
Treat agent-generated artifacts (plans, memory, configurations) as first-class data objects
Collaborate Across Pillars
Work with Models and Interfaces teams to integrate data services into training, evaluation, and inference pipelines
Partner with Infrastructure engineers to optimize data movement, tiered storage, and high-bandwidth networking (ESnet) between HPC, cloud, and edge
Engage domain scientists and mission leads for agile deterrence, energy grid, and critical materials use cases to curate problem-specific datasets
Support Continuous Acquisition & Benchmarking
Design edge-to-exascale data acquisition systems with robotics and instrument integration
Develop data/AI benchmarks—datasets, tools, and metrics—for pipeline performance, model evaluation, and mission KPIs
Author an AI-mediated parser for a new experimental instrument, automatically extracting and cataloging metadata
Implement an attribute-based policy that blocks unapproved data combinations in a classified enclave
Prototype a streaming pipeline that feeds live sensor data from a nuclear facility into an HPC training queue
Develop a dashboard that alerts on data drift, pipeline failures, or anomalous records
Collaborate with MLOps engineers to version datasets alongside model artifacts in CI/CD
Qualification
Required
Possess, or are pursuing, a PhD in Computer Science, Data Science, Statistics, or a related science or engineering field, PhD must be conferred within five years prior to employment
Experience or knowledge in these areas:
Building and maintaining production data pipelines (ETL/ELT) and data warehouses or data lakes
Programming languages such as Python, SQL, and experience with frameworks like Apache Spark or Dask
Data security and zero-trust principles, including secure enclaves, attribute-based access control, and data masking or differential privacy
Cloud platforms (AWS, Azure, or GCP) and container orchestration (Kubernetes
Ability to acquire and maintain a DOE Q-level security clearance
Preferred
Significant data research experience
Background in AI-mediated data curation: automated annotation, feature extraction, and dataset certification
Experience implementing data governance and metadata management tools (e.g., Apache Atlas, DataHub, Collibra)
Experience developing and refining data architectures and data flows
Hands-on background in MLOps and CI/CD for data and ML workflows (e.g., Jenkins, GitLab CI, MLflow)
Knowledge of human-factors engineering and UX design principles for data platforms
Knowledge of agile principles and practices and experience working as part of agile teams
Ability to work effectively in a dynamic, interdisciplinary environment, guiding technical decisions and mentoring junior staff
Strong written and verbal communication skills, with the ability to present complex data concepts to diverse audiences
Ability to obtain and maintain a SCI clearance, which may require a polygraph test
Curating and managing scientific or engineering datasets
Designing and enforcing data policies for classified, export-controlled, or proprietary data
Data architecture for HPC and edge-computing environments
Advanced data 'munging' fusion techniques for heterogeneous and streaming data sources
Building data pipelines for feature stores, experiment tracking, and model drift monitoring
Designing and enforcing data policies for classified, export-controlled, or proprietary data
Collaborating on public private partnerships or multi-lab federated data efforts
Benefits
Generous vacation
Strong medical and other benefits
Competitive 401k
Learning opportunities
Relocation assistance
Amenities aimed at creating a solid work/life balance
Company
Sandia National Laboratories
Sandia is a conducts research and development into the non-nuclear components of nuclear weapons.
Founded in 1945
10001+ employees
Funding
Current Stage
Late StageTotal Funding
$4.4MKey Investors
US Department of EnergyARPA-E
2023-09-21Grant· $0.5M
2023-07-27Grant
2023-01-10Grant· $3.7M
Leadership Team
Laura McGill
Deputy Laboratories Director - Nuclear Deterrence, and Chief Technology Officer
Maria Gallardo
CFO Enterprise Risk Management Program Lead
Recent News
Inside HPC & AI News | High-Performance Computing & Artificial Intelligence
2026-01-13
2026-01-11
Company data provided by crunchbase