Patrice Manfo Satana

– Customer need analysis: poker planning with the product owner and the business for refining the perspective and drafting the acceptance criteria for the stories committed to meeting the customer's needs – Design, develop and deploy ELT+machine Learning pipeline (DATE processing flows) and rest API to expose to external buissiness Unit – Acceptance and integration test – Delivery management

– Worked on the migration of a SAS solution to a BIG DATA solution (Spark) for monthly forecasting of the stock of assets for an investment fund. – Customer need analysis – Design and specification – Development – Acceptance and integration test – Delivery management

– Design and development of a fraud detection solution for a banking client (Cloud computing, Machine Leaning, Statistics, Data Processing) – Design and development of a chatbot telcommunication compagny (LLM : Transformer, NLU, NLG) – Development of a deep marketing engine (cloud architecture: AWS, LLM, Deep Learning, NLP) to analyze customer feedback in real time – Design and development of a product recommendation system (cloud architecture: AWS, Recommender System, Data Processing) for a company in the luxury sector – Design and development of a semantic search engine (LLM, RAG cloud architect devOps: AWS, RAG, Deep Learning, NLP) – Development of a vehicle assembly defect detection solution (Object Detection, Convolutional Neural Network, Computer Vision) for a car manufacturer

– Development of 15 Capsules (Bixby service) in accordance with the specifications – Collection, cleaning and transformation of French business data into a usable format (Cleaning, Tokenizing, NLP, Features Engineering, Web Crawling and Scraping) for model training – Creation of the training, testing and validation database – Training, testing, validation and submission of ASR-NLU (Deep Learning: deep Reccurent Neural network, Artificial Intelligence) models for BIXBY France – Pull Request : DevOps

– Extraction, cleaning and transformation of the data collected by the sensors installed on the production line and stored in the line's database – Design, development, testing, validation and commissioning of a module to automatically create prediction models – Development, testing, validation and deployment of a machine learning API (SVM, Random Forest, Naive bayse, Xgboost, PCA, Multi Gaussian) for the prediction of scale defects embedded on steel coils at the end of production – Consulting mission for Fives group's partners