Explainable AI - Postdoctoral Researcher
Core
Develop and evaluate methods for interpreting the internal representations of deep models, including sparse decompositions, concept extraction, and human-in-the-loop workflows for domain experts.
Role type
Postdoctoral Researcher (Explainable AI)
Builds
Interactive interfaces and evaluation methodologies for interrogating model internals in multimodal SciML models and deep surrogates.
Domain
National security applications, scientific computing, machine intelligence
Deliverable
production ML models | research
Required skills
Explainable AI, representation learning, mechanistic interpretability, concept-based explanation, deep learning (PyTorch/JAX), scientific programming (Python), research methodology
Preferred skills
Sparse autoencoders, transcoders, feature-learning methods, model intervention/steering, uncertainty quantification, high-performance computing, GPU programming, scientific domain knowledge (material/climate science)
Technologies
PyTorch, JAX, Python
Responsibilities
Develop methods for interpreting deep model internals; Design human-in-the-loop workflows for concept validation; Establish rigorous evaluation methodology for interpretability claims; Conduct independent cutting-edge ML research; Collaborate with project scientists on national security applications; Publish results in peer-reviewed venues.
Seniority
Postdoctoral (recent PhD holder)
