Recently graduated from the National School of Computer Science (ENSI), Tunisia, with an engineering degree in Computer Science and Artificial Intelligence. I'm passionate about Robotics, Reinforcement Learning, Embodied AI, and Robot Learning. My goal is to bridge the gap between intelligent software and the physical world by building autonomous systems that can perceive, reason, learn, and act.

• 🤖 Embodied AI & Robot Learning
• 🧠 Vision-Language-Action (VLA) Models
• 🌍 World Models for Robotics
• 🎯 Reinforcement Learning & Imitation Learning
• 👁️ Vision-Language Models & Multimodal AI
• 🗺️ Vision-and-Language Navigation
• 🦾 Autonomous Robot Control & Planning
• 🧩 LLM/VLM-based Robot Reasoning
• 🌐 Human-Robot Interaction

• 🦾 VLA & Embodied AI — exploring Vision-Language-Action models and their integration with robotic systems.
• 🌍 World Models for Robotics — studying learned environment representations for prediction, planning, and decision-making.
• 🤖 Robot Learning — experimenting with reinforcement learning and simulation-based learning for robotic manipulation and locomotion.
• 🧠 Multimodal Robot Intelligence — investigating how VLMs/LLMs can connect perception, reasoning, planning, and action.
• 📄 Research Paper — currently preparing my first research paper for journal submission.

Wilfrid Laurier University — Research Internship

Worked on Vision-and-Language Navigation (VLN) under partial observability, combining:

• Vision-Language Models for scene understanding
• LLM-based planning and reasoning
• Semantic scene graphs
• RGB-D perception
• Frontier exploration
• Deterministic robot navigation

The project explored how multimodal AI systems can improve navigation and decision-making in previously unseen environments.

AI / ML

PyTorch TensorFlow Stable-Baselines3 Gymnasium MuJoCo OpenCV Hugging Face

Robotics

ROS 2 Gazebo RViz Nav2 Isaac Sim Isaac Lab

Development

Python C++ C Git GitHub Docker Linux CUDA Jupyter TensorBoard

Research project exploring multimodal navigation using VLM perception, LLM reasoning, semantic scene graphs, frontier exploration, and robot navigation.

GPU-accelerated reinforcement learning environment for a modular quadruped robot using MuJoCo, Gymnasium, Stable-Baselines3, Docker, and TensorBoard.

Exploring the use of LLMs to automatically generate reward functions for reinforcement learning environments and robotic tasks.

Ongoing exploration of Vision-Language-Action models, learned world models, robot manipulation, and the connection between multimodal foundation models and physical agents.

Building intelligent machines that can perceive, reason, learn, and act.