Senior Data Scientist IV - AI Safety

Senior Data Scientist For Ai Safety Applied Research

The AI and Data Analytics Division, part of the National Security Directorate (NSD), combines profound domain expertise and creative integration of advanced hardware and software to deliver computational solutions that address complex data and analytic challenges. Working in multidisciplinary teams, we connect foundational research to engineering to operations, providing the tools to innovate quickly and field results faster. Our strengths are integrated across the data analytics lifecycle, from data acquisition and management to analysis and decision support.

Key Responsibilities

• Set technical direction for projects developing and applying safety methods and pipelines that analyze model internals (e.g., attention, routing, memory, tokenization, tool-use) and/or surface failure modes, capabilities, and risk profiles.
• Engage with stakeholders to translate mission needs and operational constraints into actionable safety requirements, test plans, and success criteria aligned with mission-driven goals.
• Lead projects and programs to design, implement, and validate model evaluation for cutting edge AI including frontier AI systems (e.g. LLMs, LVMs, multimodal, agentic) or development/verification of safety controls and guardrails.
• Translate cutting-edge research into mission-relevant tools and prototypes; rapidly evaluate new techniques and standards in AI safety, algorithm or system evaluation, and trustworthy AI.
• Lead development of AI safety methods across multiple axes: robustness, uncertainty quantification, manipulation, autonomy and tool-use risks, stress testing, generalization, or reliability under distribution shift.
• Build effective relationships across teams and divisions; mentor junior and senior staff; cultivate an inclusive, collaborative research culture.
• Maintain awareness of emerging trends in AI safety, AI security, alignment, national security operations, and relevant standards to shape future research directions.

Knowledge, Skills, and Abilities

• Ability to collaborate in multi-disciplinary, mission-driven teams and operate effectively in high-stakes
• Demonstrated proficiency in leading proposals, writing technical reports, and communicating complex findings to diverse audiences (technical and executive).
• Knowledge of compute environments and their cybersecurity concerns; familiarity with secure ML ops, access control, and model/data governance.
• Deep knowledge of the current ML research landscape, especially AI safety, adversarial machine learning, xAI/interpretability, uncertainty quantification, and the science of deep learning.
• Hands-on experience analyzing internal structures of deep learning models, particularly LLMs and large vision/multimodal models (tokenization, attention mechanisms, routing, weight adaptation, loss optimization) or modalities including hyperspectral imagery.
• Experience designing and executing T&E campaigns for AI systems, including test planning, dataset curation, metrics, statistical analysis, and reproducibility.
• Software engineering foundations:
  Python, ML frameworks (PyTorch/Tensor Flow), experiment tracking, and data pipeline tooling.
• Strong stakeholder engagement skills and the ability to connect technical safety work to operational mission outcomes.
• Proven track record delivering AI safety/evaluation work products in complex domains, including National Security.
• Hands-on experience with LLM/LVM/Foundation Model and Frontier AI evaluation, red-teaming, uncertainty analysis, or safety control implementation.
• Experience leading federally funded R&D projects, with publications, open-source contributions, or deployable prototypes.

Team & Culture

We foster a collaborative and innovative environment centered on continued learning, creative problem-solving, and responsible, data-driven innovation. You will work as part of interdisciplinary teams to deliver AI safety solutions for high-stakes, real-world applications. Our teams combine expertise in AI and machine learning, software engineering, human-centered design, and national security missions.