Imaging Data Scientist, ML Engineer, Computer Vision

Requirements:

• Strong Python skills and experience with scientific computing libraries such as NumPy, SciPy, OpenCV, scikit-image, and scikit-learn.
• Hands-on deep learning experience in computer vision using TensorFlow/Keras and/or PyTorch, including image segmentation and/or object detection.
• Solid machine learning engineering skills, including reproducible training, model evaluation, regression testing, and assessing model quality and failure modes.
• A disciplined approach to de-identification, consent, and data provenance.
• Helpful additional experience includes medical or scientific imaging, DICOM, physical calibration, or converting pixels into real-world measurements.
• Sports science, musculoskeletal, or physiology experience is a plus.
• Experience deploying models to constrained runtimes such as CoreML or ONNX is beneficial.
• Familiarity with labeling or annotation platforms and label round-trip pipelines is an advantage.
• Image stitching or geometric processing experience is useful.
• MLOps experience, including reproducible training, model registries, GPU scheduling, or containerized inference, is welcome.
• Experience with transformer-based or foundation-model segmentation approaches, such as SAM or medical variants, is a plus.
• Awareness of clinical software constraints, including data handling and regulatory pathways, is beneficial but not required.
• Relevant technologies include Python, TensorFlow/Keras, PyTorch, OpenCV, scikit-image, scikit-learn, pydicom, SimpleITK, cloud object storage, CoreML, Docker, Python web frameworks, and JSON-Schema-based contracts.

Responsibilities:

• Train and refine AI models that run on devices and in the cloud, and build analysis services that turn ultrasound images and clips into validated quantitative metrics for athlete performance and injury risk.
• Build image and video processing pipelines for DICOM and raw imagery, including physical calibration to produce measurements in real-world units.
• Develop musculoskeletal ultrasound segmentation models and derive athlete metrics from the resulting geometry.
• Train and package on-device navigation models that recognize views, assess image quality, and generate initial segmentations to guide image capture.
• Manage the labeling workflow by sending de-identified images for annotation, retrieving labels, and integrating them into the canonical annotation schema.
• Partner with the clinical team to define and refine labeling practices and quality standards so training data remains consistent and clinically meaningful.
• Maintain model versioning and provenance records, along with evaluation and regression-testing processes that qualify models before release.
• Observe data governance requirements by using de-identified data, respecting consent controls, and keeping identifying information out of workflows.
• Work with the mobile iOS and full-stack engineers to define shared, versioned contracts between models and the platform.
• Report directly to the CTO and take ownership across model training and production inference, with the opportunity to focus on one area according to your strengths.

Technologies:

• AI
• Cloud
• Computer Vision
• CTO
• Docker
• JSON
• Keras
• Machine Learning
• MLOps
• Mobile
• Model Training
• PyTorch
• Python
• TensorFlow
• Web
• iOS
• numpy
• opencv
• Backend

More

We are Sonari, a well-funded startup developing AI-powered ultrasound imaging for professional sports. Our technology uses portable scans to generate objective data about athletes tissue health, performance, and endurance, with the goal of predicting and preventing soft-tissue and non-contact injuries. Our approach is grounded in peer-reviewed research, and we are backed by a global investment firm with more than $70 billion in assets under management. We work with a Major League Baseball team and a Premier League club, and aim to bring this capability to sports organizations worldwide. Youll join our engineering team at the intersection of AI, medical imaging, and sports, collaborating with our Mobile iOS Engineer and Full-stack Engineer. This is a hands-on role with a broad scope; we expect to add a second ML/MLOps hire. Compensation is $160,000–$220,000 per year, with dental, health, and vision insurance. The position is hybrid remote in Cambridge, Massachusetts.

last updated 43 week of 2026