Goal: Monitored the Enedis performance & client satisfaction
Framed the business need (workshop, sketch)
Developped backend (Python | Oracle | SAS | Airflow): calculated and integrated
several KPIs
Developped and maintained a dashboard (Tableau | PowerBi)
Goal: Developped an automated summary generation model for Total
Benchmarked the best models (extractive and abstractive)
Developped and optimized the NLP summary generation model (LexRank | DistilBart)
Post-processing: developped an algorithm for recovering lost key terms
(Spacy)
Goal: Developped a tool to analyse business E-reputation thanks to press
articles
Web scrapping: Developped an automated article retrieval module
(Selenium | BeautifulSoup)
Developed a semi-supervised model to label texts: keywords detection
(KEYBERT), Word2Vec embedding, similarity calculation (cosin, Levenstein
Developped an entity detection model (NER Spacy)
Developped a classification model (Transformers XLM Roberta)
Deployed solution (Docker | Flask | Dash)
Goal: Developped an NLP model to deal with ambiguous brand names (ex:
Céline, Chloé, Total )
Data Engineering (AWS | Python | Git)
Data Analysis (Python: Pandas | MatplotLib)
Created and optimized an NLP classification model (Bi LSTM | DistilBERT)
Deployed an API model (Flack | Docker)
Goal: Developped a customer satisfaction model to evaluate the performance
of the Chatbot channel
Data Engineering
Data analysis (Python: Pandas | MatplotLib)
Created a sentiment analysis model on customer satisfaction (NLTK | ScikitLearn)
Developped and deployed a monitoring dashboard (Tableau)