Senior Data Scientist III - AI Assurance Researcher

Overview At PNNL, our core capabilities are divided among major departments that we refer to as Directorates within the Lab, focused on a specific area of scientific research or other function, with its own leadership team and dedicated budget. Our Science & Technology directorates include National Security, Earth and Biological Sciences, Physical and Computational Sciences, and Energy and Environment. In addition, we have an Environmental Molecular Sciences Laboratory, a Department of Energy, Office of Science user facility housed on the PNNL campus. The National Security Directorate (NSD) drives science-based, mission-focused solutions to take on complex, real-world threats to our nation and the world. The AI and Data Analytics Division, part of the National Security Directorate, 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. Responsibilities PNNL is seeking a deep learning researcher to investigate how AI models learn and make decisions, and to develop methods for assessing and improving their reliability. You will connect internal representations and computations to observable behavior, using those findings to explain model decisions and examine how uncertainty, changing conditions, or adversarial inputs affect the reliability of model outputs. At PNNL, your research will address national security and safety needs. You will have opportunities to collaborate across the U.S. government and develop and pursue new research ideas that advance AI Assurance and address mission needs. You will lead research projects or major parts of larger efforts, develop research questions and methods, and mentor colleagues while remaining hands-on in experimental research. Relevant experience may include research on deep learning fundamentals, such as optimization, representations, or generalization. Key Responsibilities • Lead research projects or major parts of larger efforts, coordinating the technical approach, team contributions, and research milestones. • Formulate research questions, develop or adapt methods, and design and execute deep learning experiments, including large-scale experiments on high-performance computing (HPC) infrastructure as appropriate. • Investigate how model architectures, training dynamics, and internal computations shape behavior, using findings from systems such as large language and vision models to explain observed capabilities and failure modes. • Develop methods to quantify uncertainty and assess robustness under distribution shifts or adversarial inputs, using red and blue teaming to identify vulnerabilities and evaluate defensive methods. • Develop reproducible evaluations to investigate model reliability and failure modes, compare methods against appropriate baselines, and communicate findings, uncertainty, and limitations. • Develop reproducible research code and collaborate with engineers to translate findings and prototypes into useful tools and capabilities. • Lead the preparation of publications, technical reports, and presentations that communicate the team’s findings. • Mentor researchers and engineers through technical discussions, feedback, and collaborative research. • Develop research ideas and contribute to proposals with collaborators. • Conduct work in secure environments in accordance with operational security requirements. This position is based in either Richland, WA or Seattle, WA and requires an onsite presence. Qualifications Minimum Qualifications: • BS/BA and 5+ years of data science experience -OR- • MS/MA and 3+ years of data science experience -OR- • PhD with 1+ year of data science experience • Experience conducting deep learning research to investigate how models learn, behave, or generalize. • Experience using Python and PyTorch for deep learning research. Preferred Qualifications: • Experience explaining model behavior through analysis of internal representations and computations, including mechanistic interpretability of large language or vision models. • Experience developing or evaluating methods to assess tool-using models and multi-step agentic workflows, including analysis of execution traces, task outcomes, and failure modes. • Experience designing and executing large-scale deep learning experiments. • Experience leading research projects or major technical efforts, including formulating research questions, developing or adapting methods, coordinating contributors, and evaluating and communicating results. • Experience translating research prototypes into tools or capabilities or contributing peer-reviewed publications or other technical contributions in relevant research areas. • Experience supporting national security missions. Hazardous Working Conditions/Environment Not Applicable Additional Information This position requires the ability to obtain and maintain a federal security clearance. A security clearance background investigation includes review of your employment, education, financial, and criminal history, as well as interviews with you and your personal references, neighbors, and co-workers to determine trustworthiness, reliability, and loyalty to the United States. The investigation also examines your foreign connections, drug and alcohol use, foreign influence, and overall conduct. Requirements: • U.S. Citizenship • Background Investigation: Applicants selected will be subject to a Federal background investigation and must meet eligibility requirements for access to classified matter in accordance with 10 CFR 710, Appendix B. • Drug Testing: All Security Clearance po