Mohamed Z.

• Development of a web application with Laravel and MySQL.
• Design of an intelligent chatbot based on a RAG (Retrieval-Augmented Generation) architecture.
• Implementation of semantic search with BGE-M3 and Qdrant.
• Development of an automatic sentiment analysis system with XLM-RoBERTa.
• Automatic topic detection using BERTopic.
• Generation of intelligent recommendations and creation of a decision support dashboard.
• Integration of Artificial Intelligence models for data analysis and exploitation.

Missions performed

Predictive modeling:

• Use of Facebook Prophet to model and predict production evolution
• Parameter optimization to improve accuracy (MAE, RMSE, R²).
• Generation of annual forecasts until 2040.

User interface development:

• Creation of an intuitive Streamlit application with multi-page navigation (History, Forecasts, Home).
• Integration of interactive filters (product, year range).
• Display of key statistics (average, min, max, growth).
• Graphical visualization with Matplotlib and export to CSV/PNG format.
• Comparison and analysis :Automatic calculation of the variation between the last historical value and the first forecast to measure the expected impact.
• Deployment :Online deployment of the application on Streamlit Cloud.

Technologies used

Results & added value

• Interactive tool usable by agricultural professionals, analysts, or public decision-makers.
• Clear visualization of past and future trends.
• Easy export of data and graphics for integration into reports.