En tant que CTO et co-fondateur d'Edzo, je suis responsable de la stratégie technologique de l'entreprise et de la gestion des aspects techniques de l'entreprise. Mon rôle consiste notamment à superviser l'utilisation des logiciels et des outils pour la gestion des projets, ainsi qu'à veiller à ce que ces systèmes fonctionnent de manière efficace et sécurisée.En tant que co-fondateur, je suis également chargé de diriger l'équipe technologique pour mener à bien les différents projets dans les délais et les budgets prévus. Je suis également impliqué dans la création de nouvelles solutions technologiques pour l'entreprise, comme un outil de gestion des tâches en temps réel, un système de recherche sémantique textuelle ou un outil de suivi des performances des équipes.En somme, je suis un expert en technologie et je m'assure que notre société dispose des outils nécessaires pour atteindre ses objectifs.

■ Article classification:I worked on deep learning methods to find among millions of articles, the articles that could be the most interesting for a customer using language models and words embedding.Keywords : NLP, Keras, LSTM, ELMo, deep learning■ Find the best sentence and replace the handwritten system :I set up an intelligent system to find the most attractive sentences in an article and use it as a post with automatic hashtags when the article is shared on different social networks.This intelligent system has made it possible to replace handwriting sentences with an automatic system used in more than 70% of cases with a great time saving.Keywords : NLP, Keras, LSTM, deep learning■ Twitter accounts recovery:I carried out the recovery of Twitter accounts from cities with more than 10,000 inhabitants to broadcast local content.Keywords : NLP, regex, Twitter API■ Articles exclusion:I implemented an solution to exclude articles containing prohibited words using Aho–Corasick algorithm.Keywords : NLP

The purpose of this project was to estimate if the artists mentioned in the same text could help to recommend music on the platform.■ Named Entity Recognition:Develop a system that performs Named Entity Recognition (NER), allowing to detect mention of musical entities (eg. artist names) in corpus of texts. Scores are calculated on freshly new datasets.Keywords: NLP, NER, Tensorflow, BI-LSTM-CRF, spacy, word2vec■ Creation of new datasets exploiting the internal redirections of different social platforms: ●● Twitter: - Recover artists' Twitter account using the tweets posted by the account (tf-idf and SVM)- Replace artists' account names with their real names in tweets to automatically annotate tweets (NLP)541661 tweets automatically annotated. ●● Facebook: - Recover artists' Facebook account using the artists' Twitter account name (NLP, regex, Facebook API)- Replace artists' account names in posts with their real names to automatically annotate posts (NLP)13056 Facebook's posts automatically annotated. ●● Wikipedia: - Retrieving Wikipedia pages with a music theme and then searching through each link on each page for the name of an artist (NLP, DBpedia, SVM).- Replace artists' wikipedia links in pages with their real names to automatically annotate Wikipedia page (NLP).104725 automatically Wikipedia pages annotated.

■ Named Entity Recognition: Develop a system that performs Named Entity Recognition (NER) and anaphora/coreference resolution, allowing to extract sources cited in a journalistic article (eg. political figure) using machine learning (notably the Conditional Random Fields (CRF)).The project was in partnership with Le Monde and the AFP (Agence France Presse) and is used for their services.