Post Doctoral Scholar

The Ohio State University • Columbus, OH

2d • Onsite

About The Position

Postdoctoral Scholar in 3D Artificial Intelligence for Biological Imaging
The National Science Foundation (NSF) Imageomics Institute seeks a postdoctoral researcher to develop and apply artificial intelligence methods for three-dimensional biological image data. The position is for one year.
Biological imaging increasingly produces volumetric and surface data at a scale that manual analysis cannot match: tomographic volumes, surface meshes, point clouds, and derived geometric measurements across hundreds or thousands of specimens. The successful candidate will build methods that turn these data into quantitative biological measurements and will apply those methods to open scientific questions in organismal and evolutionary biology.

Requirements

• PhD (completed or expected before the start date) in computer science, biomedical engineering, applied mathematics, statistics, a quantitative biological discipline, or a related field.
• Demonstrated research experience in machine learning or computer vision, evidenced by publications, preprints, or a public code record.
• Strong programming ability in Python, and practical experience with a modern deep learning framework such as PyTorch.
• Experience working with 3D data in some form: volumetric images, meshes, point clouds, or geometric representations.
• Ability to work independently on an open-ended research problem and to communicate results clearly in writing and in talks.

Nice To Haves

• Experience with segmentation of volumetric image data, and with the practical difficulties of annotation, class imbalance, and small training sets.
• Familiarity with foundation models and with methods for adapting them to new domains, including fine-tuning, prompting, and parameter-efficient approaches.
• Background in geometric deep learning, shape analysis, or statistical shape modeling.
• Experience with vision-language or other multimodal models.
• Prior work with biological or biomedical imaging data, and comfort collaborating with domain scientists.
• Experience running work on high-performance or cloud computing resources.
• A record of contributing to open-source scientific software.

Responsibilities

• The work centers on 3D segmentation and the adaptation of foundation models to volumetric biological data.
• Evaluating and adapting promptable and automated segmentation models for volumetric biological imaging, including cases where the models were trained on data quite different from ours.
• Developing methods for initializing and propagating segmentations across specimens, such as atlas- or template-based approaches, and quantifying how well they perform against expert annotation.
• Designing evaluation protocols that report method performance honestly, including where and why methods fail.
• 3D shape analysis and geometric machine learning. Learning on meshes, point clouds, and landmark configurations; automated landmark placement; establishing correspondence across specimens; and scaling geometric morphometric analysis to large collections.
• Multimodal and generative approaches to 3D biological data. Connecting 3D structure to text, images, and other modalities; retrieval and description of 3D specimens; and generative models of biological shape.
• Across all of these, the candidate will be expected to release working code and to document methods so that other researchers can reproduce and build on them.

Benefits

• excellent benefits

What This Job Offers

• Job Type: Full-time
• Career Level: Entry Level
• Education Level: Ph.D. or professional degree