Conversational AI Agent for learning chess openings (RAG, LangGraph, MCP)
Conversational AI Agent combining RAG, chess engine, and LangGraph to teach openings to young talents of the Fédération Française des Échecs.
Conversational AI Agent combining RAG, chess engine, and LangGraph to teach openings to young talents of the Fédération Française des Échecs.
SFT + DPO fine-tuning of an LLM (Qwen3-1.7B) to assist medical triage in emergency rooms, with particular attention to GDPR compliance.
Machine Learning model predicting the risk of resignation of employees of an ESN from RH data and satisfaction surveys.
Complete strategic framing of an AI project for clothing recommendation by photo, from product idea to costed proposal for a COMEX.
Application of Data project management methods to the structuring and steering of my own professional AI Engineer portfolio.
Credit scoring model estimating loan default risk, with experiment tracking and version management via MLflow.
Deployment to production and monitoring of a scoring model via a containerized and automatically deployed API (MLOps).
Industrialization of a Machine Learning model via a secured, tested, and continuously deployed FastAPI REST API.
Complete progression in Reinforcement Learning, from Q-Learning up to autonomous control of a lunar lander exposed via an API.
Weak labeling pipeline and semi-supervised classification model for brain tumor detection from sparsely annotated MRI.
Industrialized ETL Pipeline extracting text + image news data to feed a fake news detection engine.
Machine Learning model predicting the energy consumption of non-residential buildings in Seattle based on their structural characteristics.
Prototype of automatic clothing segmentation on images via a vision model hosted on Hugging Face Serverless endpoints.
POC of a RAG system answering in natural language questions about cultural events, combining FAISS vector search and LLM.