Argonne National Laboratory is seeking to advance trustworthy, privacy-preserving AI methodologies that can safely leverage sensitive scientific data, particularly in biomedical applications. The goal is to develop next-generation Privacy-Preserving Federated Learning (PPFL) techniques to enable collaborative model training without centralized data movement.
Requirements
- Demonstrated experience in programming, with proficiency in Python and familiarity with numerical or scientific computing libraries (e.g., NumPy, PyTorch, TensorFlow).
- Strong aptitude for developing and evaluating machine learning models, including hands-on experience implementing algorithms for classification, regression, or representation learning.
- Ability to analyze complex datasets, design computational experiments, and interpret model performance in a scientifically rigorous manner.
- Experience with privacy-preserving machine learning concepts such as federated learning, differential privacy, or secure multiparty computation.
- Familiarity with handling biomedical or multimodal data (e.g., images, EHR, genomics, sensor data).
- Exposure to large-scale computing environments or distributed systems concepts.
- Understanding of deep learning architectures (transformers, CNNs, RNNs) and approaches for multimodal fusion.
Responsibilities
- Develop and evaluate PPFL algorithms that enable collaborative model training on distributed biomedical datasets without centralized data movement, leveraging frameworks such as APPFL/APPFLx and secure HPC environments.
- Create novel approaches for multimodal learning in federated settings, integrating structured EHR data, imaging, sensor data, and genomic information to improve model robustness, generalization, and interpretability.
- Study privacy budgets and differential privacy mechanisms to quantify, optimize, and communicate tradeoffs between privacy guarantees and downstream model performance, with emphasis on biomedical use cases.
- Design and run large-scale experiments on DOE leadership-class computing systems, benchmarking federated optimization strategies, privacy mechanisms, and algorithmic scalability.
- Contribute to scientific discovery in computational biology and precision medicine, including applications such as disease risk prediction, phenotype modeling, digital pathology, and clinical decision support.
- Disseminate research findings through peer-reviewed publications, technical reports, conference presentations, and open-source software contributions.
- Collaborate closely across Argonne’s multidisciplinary teams, including experts in HPC systems, AI frameworks, privacy-enhancing technologies, and biomedical data science, in alignment with Argonne’s strategic mission in computation and health-related AI.
Other
- Recently completed Master’s degree in computer science, biomedical engineering, applied mathematics, statistics, computational biology, or a related quantitative field.
- Excellent written and verbal communication skills, with the ability to work effectively in interdisciplinary research teams.
- Ability to model Argonne's core values of impact, safety, respect, integrity and teamwork.
- Willingness to disclose participation in Foreign Government Sponsored or Affiliated Activities as per United States Department of Energy Order 486.1A.
- Ability to pass a background check that includes an assessment of criminal conviction history conducted on an individualized and case-by-case basis.
