APPRENTISSAGE - DATA SCIENTIST (H/F)

Contenu de l'offre APPRENTISSAGE - DATA SCIENTIST (H/F) chez Ipsos

À propos

Ipsos est le leader en France des études de marché et des enquêtes d’opinion. Groupe français indépendant, fondé en 1975, Ipsos opère aujourd’hui dans 90 pays et donne aux entreprises et aux institutions pour lesquelles il travaille, les clés de compréhension de la société, des marchés, des citoyens et des consommateurs.

Une mission qui s’appuie sur la disruption numérique, l’extension des domaines d’observations et l’innovation continue pour améliorer la finesse des analyses.

Intégrer Ipsos aujourd’hui c’est :

Choisir un métier passionnant au cœur des enjeux d’actualités et des transformations de la société. Bénéficier d’une multiplicité d'opportunités d’apprentissage dans des secteurs différents : le marketing, la publicité, les médias, la politique, le CRM, qui intègrent les dernières innovations digitales et technologiques. Se voir confier des responsabilités immédiatement sur des études avec de véritables perspectives d’évolutions professionnelles. Evoluer au sein d’un groupe international dans lequel les opportunités de mobilité sont une réalité.

Pour répondre aux nouveaux besoins de ses clients, Ipsos dispose de plus de 75 services différents regroupés dans 17 Lignes de Services.

Cette nouvelle organisation nous permet d’aiguiser nos expertises pour apporter à nos clients des réponses précises à leurs questions, en nous permettant d’être plus spécialisés, rapides et homogènes sur tous les marchés.

Descriptif du poste

Synthesio, an Ipsos company, is a global leader in AI-enabled Consumer Intelligence. Our hybrid offering provides companies, brands, and agencies with the most complete, accurate, and predictive picture of their markets and buyers. Our AICI platform, powered by the most advanced natural language understanding and AI algorithms, supports the broadest set of online and offline data sources and fully
leverages Ipsos’ award-winning analytical frameworks.
Synthesio was founded in2006 and has offices in New York, Paris, London, Singapore, and Brussels.

The Synthesio R&D department is focused on data provisioning, enrichment and exploitation.
The crawl 70M posts each day from a plethora of social networks and mainstream press websites.
Each document is then analysed and enriched using custom NLPand Computer Vision models developed by our data scientists. All this data is stored in our databases (1 PB+ ElasticSearch, 750TB+ MySQL clusters, 50 TB+ ScyllaDB) and can be accessed by our customers via our dashboarding solution.

Our stack: Go, Python, NodeJS, MySQL, ElasticSearch, Cassandra (ScyllaDB),Kafka, Gitlab, Docker, Ansible.

We are looking for a junior data scientist to join our R&D team in Synthesio’s Paris office for an alternance contract on the theme of deep learning for logo recognition.

Context: logo recognition is an essential part of our computer vision capabilities. It allows our clients to track their visual brand presence outside the realm of explicit textual mentions.
Our logo recognition system is currently capable of detecting the logos of about 700 brands.
We are continuously working to increase the number of logos the system can recognize to keep up with the demands of our clients. To do this, we need to find example images, annotate them with bounding box annotations, augment with synthetic images, retrain the logo recognition model on the new data, and finally deploy the model in production.

We are also interested in finding ways to improve the accuracy of the model and its sample efficiency i.e. the number of images needed to reach a reasonable average precision for newly supported logo classes.

Concretely, your mission will be to:

Share ownership of our logo recognition model with the data science team Create and label training images for supporting new logos Improve our tooling and the level of automation around image labeling and model retraining Run training experiments for logo recognition Deploy the logo recognition model in production Research ways to improve the accuracy and sample efficiency of logo recognition

Profil recherché

Technical Skills :

Solid understanding of machine learning and deep learning Solid understanding of the fundamentals of algorithms and data structures Proficiency in Python Familiarity with the essential machine learning toolkit (scikit-learn, PyTorch, Tensorflow, Keras, etc.) University/school projects involving deep learning for NLP/Computer Vision are appreciated

Soft Skills :

Ability to work in a multicultural environment Good communicator, self-starter and collaborative enthusiast Rigour and strong appetite for software quality Fluent in English and French. This includes written, speaking and reading

Process de recrutement

A 30 minute screening phone call with the hiring Engineering Manager A technical test to assess your data science skills A 60 minute interview to discuss your solution to the test (video call possible)

Informations complémentaires

Type de contrat : Alternance (6 à 12 mois) Date de début : 01 février 2022 Lieu : Paris, France (75013) Niveau d'études : Bac +5 / Master Expérience : < 6 mois