STAGE Amélioration des algorithmes d'Object Detection

Gleamer develops state-of-the-art Artificial Intelligence algorithms to make a transformative impact
in healthcare.
Our first product is an AI-based diagnosis workstation for radiologists to improve their reliability and
productivity. We use cutting-edge deep learning algorithms to help radiologists quickly detecting
abnormalities in X-rays with very high sensitivity. Our algorithms aim to prevent most of missed
lesions and pathologies while permitting radiologists to spend more time with their patients.
At Gleamer, you’ll find a dynamic team with a strong company culture. We believe that our team will
be the major asset to achieve the profound challenges we address. Alongside this, we have raised
1,5M€ with top-tier investors, including Elaia Partners and XAnge. We also benefit from the
stimulating incubator Agoranov in the heart of Paris.
We are looking for a passionate and talented AI Scientist with strong expertise in deep learning
applied to computer vision. Most critically, we believe that an acute critical spirit is an imperative, as
far as you’ll be working on data and algorithms that will eventually impact millions of people. We lay
great importance on the ability to communicate and being independent, as you will be implied in a
quickly growing structure, working closely with research engineers, doctoral researcher, developers,
and medical doctors.
In this research-oriented internship, the intern will strive to enhance the existing multitask learning
pipeline. During the application process, discussions with the candidate should result in the
establishment of one or several supervision signals that could improve the anomaly detection model.
Once settled on the nature of said signals, Gleamer will create a dataset with the corresponding
annotations, so that the intern can experiment quickly with his ideas upon his arrival. Additionally,
the intern will also be able to play with the metadata that comes with the considerable dataset of
annotated radiographs that our anomaly detection model is trained with.