Data Science intern

Each year, 140M children are born. Birth defects affect 1 in 33 births in developed countries, and in 50% of cases, they are not detected during prenatal examinations. Prenatal medicine is particularly complex, and the scans impose heavy responsibilities on healthcare professionals. They can also be a source of stress for future parents.

Resulting from 5 years of collaboration between world-renowned experts from Necker Hospital and Ecole Polytechnique, Sonio uses artificial intelligence to improve prenatal screening and diagnosis. Based on patented algorithms and a proprietary expert database, Sonio aims to become the reference tool to help practitioners improve screening, reduce diagnostic errors, and optimize pregnant women’s medical path.

Sonio's mission is to improve women's and children's health by promoting access and quality of care through technological innovation, medical expertise, and collective intelligence.

About the role

We are seeking a highly motivated and talented intern to contribute to a project focused on the development of a deep learning foundation model for ultrasound imaging. The goal is to build a versatile deep learning model that can be fine-tuned and applied across a variety of downstream tasks, with a particular focus on medical image segmentation.

The intern will work with our data science team to design, train, and evaluate this model. This project will involve a deep understanding of foundation models, transfer learning, and segmentation techniques. It will provide valuable hands-on experience in applying machine learning to real-world healthcare problems.

• Model Development: Implement a deep learning foundation model, with architectures like CNNs, Transformers, or other state-of-the-art networks suitable for ultrasound data.

• Transfer Learning: Investigate the use of pre-trained models and apply transfer learning strategies to optimize model performance on specific ultrasound segmentation tasks.

• Model Evaluation: Design and execute rigorous testing and evaluation procedures to measure model performance, especially in segmentation accuracy. Apply metrics such as Dice Score, Intersection over Union (IoU), and other key indicators for medical image quality.

• Benchmarking: Compare the foundation model’s performance against existing models for ultrasound segmentation tasks.

• Collaboration: Work with domain experts to understand clinical applications, validate models, and ensure that the system meets healthcare standards.

About the profil

• Educational Background: Currently pursuing a master’s degree in Computer Science

• Technical Skills:

  • Strong knowledge of Pytorch

  • Experience with image processing and computer vision tasks.

  • Familiarity with medical imaging is a plus.

  • Proficiency in Python and data handling libraries

  • Proficiency in SQL

• Research Mindset: Strong interest in applied research and innovation in healthcare AI.

Location : Paris

We move fast and aspire to be transparent over the process - our objective is that the process from the first chat to an offer is no longer than a month. We also aspire to give an answer to every application in a week - if you have not heard from us, please follow up at careers@sonio.ai.