Researcher

ages, intraoperative video, OR traffic signals, EHR data, and patient/surgeon surveys with explainability and equity at its core. We seek a Computer Science Postdoctoral Associate (Postdoc) to lead vision–text transformer modeling and interpretable, clinically aligned AI for this multimodal platform, working alongside surgeons, infectious disease experts, and ML faculty. In addition to Rutgers-based mentorship, the Postdoc will collaborate closely with lead computer scientists across our partner institutions—including Stanford, the University of Florida, and Weill Cornell Medicine—along with other academic and industry collaborators through co-mentorship, cross-institution model reviews, and shared code/MLOps standards. The Postdoc is expected to maintain a high level of communication with the PI and primary mentorship team as described in the BMIHAI PAIR grant.. . Candidates must hold a PhD in Computer Science, Electrical/Computer Engineering, or a related field by the start date, with a strong publication record in computer vision, multimodal learning, or vision–language models. We require hands-on expertise with transformer architectures (e.g., ViT/TimeSformer, CLIP/BLIP or similar) in PyTorch, including scalable training on GPUs and reproducible experimentation. Demonstrated experience building explainable models (e.g., concept bottlenecks, prototype/contrastive methods, attribution) and evaluating calibration and uncertainty is essential. Strong software engineering skills—Python, Git, unit testing, experiment tracking—and fluency with containerization (Docker) and basic cloud workflows are expected, alongside excellent communication and cross-disciplinary collaboration with clinical teams.. . The Postdoc will participate in lab meetings, research in progress seminars, and presentations at conferences. They will prepare manuscripts based on their research, aid in developing the PI’s grant applications, and will be expected to apply for personal grant funding. Multiple resources for career development to aid with these requirements will be provided by the PI and the Office of Postdoctoral Affairs.. . Among the key duties of this position are the following:. - Design and train vision–text transformer architectures for multimodal fusion (RGB + thermal, intraoperative video, OR signals, EHR, surveys); develop temporal and cross-attention components for longitudinal risk trajectories.. - Build robust fairness and skin-tone–aware pipelines (Fitzpatrick spectrum robustness, augmentation, bias audits).. - Partner with annotation teams using Encord (video) and our SegmentCare image labels; shape active-learning loops and QC.. - Productionize models with PyTorch, Docker/Kubernetes, and AWS/SageMaker. - Ensure all members of the lab are in compliance with the necessary trainings (both at the lab and institutional level).. - Keep up to date on the relevant literature in both peer-reviewed and preprint form.

Experience: Required
Employment: Full-time
Salary: $50,000 – $120,000 yearly

About Rutgers Robert Wood Johnson Medical School:

Rutgers Robert Wood Johnson Medical School is one of the nation’s leading comprehensive medical schools, committed to excellence in education, research, healthcare delivery, and community health promotion. As part of New Jersey's premier ac