Cette année nous ouvrons trois postes de stage chez nam.R au sein de l’équipe Data science.
Toutes nos offres de stage sont encadrées par deux tuteurs et le stagiaire est intégré au sein des équipes Data. Le sujet est défini en amont du stage et reste la préoccupation principale de l’étudiant.

sujet restant :

• OFST2021_CV2: Active learning implementation for computer vision
  Nam.R is using state-of-the art machine learning and deep learning methods to build its Digital Twin, a (BIG) geospatial, structured database in France containing information about territories, buildings, cities and more.
  Our computer vision engineers design deep learning pipelines to extract meaningful information from street, aerial or satellite imagery. To train their models, they get tagged images by in-house human annotators. This process is time consuming and expensive to obtain, but essential for CV models.
  In order to improve this process, we would like to set up an Active Learning (AL) approach. The idea of AL is, instead of just giving the learner a lot of data to learn from, to allow the learner to ask questions about the given data. In particular, the learner gets to ask an oracle (human annotator) about the label of certain instances that are currently unlabeled. If the learner asks smart questions, he might be able to get examples that are very informative and reach a high level of generalization accuracy with a much smaller labeled dataset than he would have if that dataset had been created using random sampling.

Formation : Master 2 ou Ecole d’ingénieurs en Data Science / Statistiques / Mathématiques / Mathématiques Appliquées

Vous avez :

• une connaissance des algorithmes phares du machine learning en supervisé (classification, régression, etc) et non supervisé (clustering, feature extraction/selection, etc)
• une bonne maitrise des langages Python (scikit-learn, matplotlib, keras, etc) et SQL
• des connaissance des outils PostGIS, QGIS, Dataiku, Git ainsi que de l’écosystème Big Data (Hadoop, framework Spark et/ou Splunk) serait un plus

1/ Candidature avec CV et conditions de stage : date, durée, année de validation..
2/ Test quizz
3/ Entretien