Rayen
Ammar
AI & Data Science Student | AI/ML & Full Stack
Building end-to-end AI systems and data-driven platforms, from predictive ML models to production full-stack applications.
About Me
I build AI-powered systems and full-stack applications from ML pipelines to production web platforms.
I'm a third-year AI & Data Science engineering student at TEK-UP University, Tunisia. I work across the full spectrum of AI/ML and full-stack development, building tools that turn complex data into decisions that matter.
My projects span workforce intelligence (CHRONOS), ocean analytics (waveAI), financial AI (DUALIS), NLP recruitment (UNIVIOO), and retail forecasting (RETALIS). I'm picking the right tool for each problem: Python + PyTorch for AI/ML, Django or Spring Boot for backends, React for frontends.
I'm actively seeking an AI engineering internship where I can contribute to impactful, data-driven systems.
Technologies
Python
PyTorch
Scikit-learn
React
Spring Boot
Django
PostgreSQL
MongoDB
Docker
Git
JavaScript
Azure
LangChain
Neo4j
FastAPIProjects
Global ocean intelligence platform using LSTM autoencoders and Random Forest for storm detection, safe departure window prediction, and optimal fishing zone identification. SHAP explainability applied across satellite and oceanographic data sources.
Designing an evidence-grounded agentic system to analyze potential drug interactions, contraindications, and medication risks in patient context. The workflow combines drug normalization, graph-based relationships, vector retrieval, multi-source evidence gathering, and a verification step that flags insufficient or conflicting evidence rather than forcing a definitive answer.
Designed a two-lane support architecture for internal IT and external customer service. AI capabilities include ticket classification and routing, RAG-grounded response suggestions with supporting evidence, and anomaly detection for unusual ticket-volume spikes. The workflow keeps human agents in control and outlines experiment tracking, model versioning, monitoring, and retraining.
End-to-end time series analysis on multi-source retail data. Temporal features engineered (paydays, weekends, holiday flags) and fed into LightGBM and XGBoost models with TimeSeriesSplit cross-validation. Interactive dashboards revealed seasonal trends and macroeconomic correlations.
AI-driven lending platform automating loan approval and financial advisory. Predictive ML model assesses applicant eligibility from financial profiles; generative AI produces transparent, human-readable explanations for each lending decision.
AI platform automating CV parsing and candidate–opportunity matching via TF-IDF cosine similarity. A generative AI chatbot delivers personalized career guidance within a modular multi-stakeholder system for candidates, employers, and administrators.
Lightweight CRM with AI-generated priority classification (High / Medium / Low), engagement summaries, and keyword extraction from notes and CVs via OpenAI. Dashboard surfaces AI-driven insights for actionable client decision-making.