AI Engineering Student · Researcher · Builder
I'm Aya, a fourth-year AI & Data Science Engineering student at the National Higher School of Artificial Intelligence (ENSIA) — one of Algeria's most selective institutions dedicated entirely to AI. I enjoy exploring different areas of artificial intelligence through coursework, research projects, internships, and personal experiments.
My interests span a wide range of topics across artificial intelligence, from reinforcement learning and self supervised learning to transformers, generative models, and representation learning. Exploring different areas has helped me appreciate how ideas developed in one field often inspire progress in another.
What excites me most, however, is understanding why things work rather than simply using them. I'm particularly interested in the theoretical foundations of machine learning—studying the assumptions, mathematical principles, and design choices behind algorithms, and investigating the mechanisms that make them effective. I enjoy reading papers, experimenting with ideas, and building projects as a way to bridge theory and practice, with the hope of eventually contributing original insights of my own.
- Reinforcement Learning for autonomous systems
- Self-Supervised Learning & JEPA world models
- Scientific Machine Learning
- Natural Language Processing for low-resource languages
- Privacy & Security in AI
- Graph Representation Learning
- Remote Sensing & GeoAI
- Agentic AI systems with memory and orchestration
PyTorch · TensorFlow · Keras · Scikit-learn · Stable-Baselines3 · Gym · NLTK · OpenCV · Pandas · NumPy · Matplotlib · Seaborn · Jupyter · Docker · Flask · Git/GitHub · Linux · Deep Learning · Reinforcement Learning · NLP · Computer Vision · Time Series Analysis · Data Mining · Optimization · GeoAI & Remote Sensing · Full-Stack Development
| Project | Description | Stack |
|---|---|---|
| Autonomous Drone Delivery via Reinforcement Learning | Curriculum-driven PPO framework for autonomous multi-client drone delivery under wind disturbances, obstacle avoidance, and energy constraints, achieving a 96% success rate in PyBullet | PyTorch, Stable-Baselines3, PyBullet |
| Human-in-the-Loop Arabic Machine Translation | Domain-aware Arabic MT framework integrating LLMs, RAG, and RLHF to improve translation quality, terminology consistency, and reduce hallucinations | PyTorch, RAG, RLHF, Qdrant |
| Physics-Aware Transformer for PDE Solving | Hybrid Fourier Neural Operator–Transformer with physics-informed losses for autoregressive PDE forecasting, outperforming PINNs under out-of-distribution conditions | PyTorch |
"إذا قلت قد علمت فقد جهلت"