AI & Machine Learning Student | Deep Learning Enthusiast
Here are my academic projects and professional experiences, showcasing my skills in programming, AI, Machine Learning and Deep Learning.
This applied research project, conducted at the GREYC Laboratory, sought to reimagine textile art using technology. By leveraging generative diffusion models, the project successfully created original-style tapestry pieces and explored the structuring of visual narrative sequences within the generated imagery.
A four-month internship within the Department of Robotics and Mathematics provided a comprehensive introduction to Reinforcement Learning (RL) under the direct supervision of university researchers and professors. The key technical achievement involved applying core RL concepts to simulated robot locomotion. This included designing the necessary observation and reward functions to successfully model the movement of the Go2 quadruped robot.
Developing a reinforcement learning agent using Deep Q-Network (DQN) to solve grid-based pathfinding problems. The agent learns to navigate from a starting position to a goal while avoiding randomly generated walls in a 5x5 grid environment. The project implements experience replay, target networks, epsilon-greedy exploration, and cycle detection mechanisms to improve learning stability and efficiency.
Developed an intelligent conversational chatbot for LCL Bank during a highly competitive 26-hour hackathon in Paris. This solution utilized Dialogflow for natural language processing and Retrieval-Augmented Generation (RAG), integrated with Google Cloud Platform, to deliver fast, advanced customer support and banking assistance.
Developed along with a team of 4, a complete tournament management system for the virtual Blokus game to address and automate the organizational complexities of future ENSICAEN programming challenges. This robust platform centralizes all tournament organization, eliminating the time-consuming manual management of scores and matches. The system's utility is further enhanced by the inclusion of integrated AI opponents for testing and development.