Wassim Ayari

Observatory of Fundamental Rights – AI Legal Platform – Developed a platform to centralize and analyze case law related to fundamental rights in Tunisia. – Designed an AI module capable of analyzing and understanding legal texts. – Built a contextual and semantic search engine using AI to assist users in legal research – Created a virtual assistant to guide non-lawyers on the platform. Technologies: Python, Transformers (GPT), LangChain, RAG, LLM fine-tuning, FastAPI, PostgreSQL, Vector Databases (FAISS), Docker HotelGenuis.App – AI Concierge Platform – Multilingual AI assistant for hotel guests (recommendations, bookings, instant support). – Real-time monitoring of customer requests and AI responses (API + dashboards). – Predictive ML models to anticipate guest needs and increase upselling. Technologies: Python, TensorFlow, Scikit-learn, PostgreSQL, Redis, FASTAPI, Predictive Modeling Attijari Bank - AI-powered Internal Assistant for Ticket Management & Knowledge Automation – Designed and deployed an AI-driven internal assistant to automate ticket processing for bank employees, improving response time and internal workflow efficiency across departments. – Built an intelligent system capable of understanding and answering internal circulars, memos, and financial procedures, – Integrated the assistant with the bank's internal ticketing system and enabling automatic classification. – Implemented intern LLM and document understanding models to analyze regulatory documents, financial processes, and internal policies. Technologies: Python, Transformers, LangChain, RAG, FastAPI, PostgreSQL, Vector Databases (FAISS / Pinecone), Docker.

– Production deployment of product classification models for imbalanced datasets using deep neural networks. – Built production-ready ML pipelines with CI/CD integration. – Document classification using Transformers, LangChain, and generative AI. – End-to-end MLOps for retail and e-commerce platforms. Technologies: Python, Scikit-learn, TensorFlow (Keras), Transformers (BERT, GPT), LangChain, FastAPI, Vertex AI, BigQuery, MongoDB, GitLab CI/CD, Docker.

– Designed CNN and PointNet neural networks for 2D and 3D image classification – Developed ML models such as XGBoost and RandomForest. – Preprocessed imagery and structured datasets. Technologies: Python, TensorFlow, PyTorch, CNN, PointNet, Scikit-learn, PostgreSQL, REST APIs

– Built predictive ML models (XGBoost) for anomaly correction. – Created postal address correction algorithms using ML and rule-based systems. – Data preprocessing and feature engineering. Technologies: Python, XGBoost, TensorFlow, Scikit-learn, REST API, PostgreSQL

– Geospatial data analysis to detect event conflicts. – Event risk prediction using RandomForest and XGBoost. – Generated KPIs for city operations. Technologies: Python, PySpark, Hive, Hadoop (Cloudera), MongoDB, Scikit-learn, GIS data processing

– Deployed large-scale recommendation engines for Carrefour loyalty customers. – Computed affinity scores and built navigation KPIs. – Ran workshops with marketing and data teams. Technologies: Python, PySpark, Hive, Hadoop, BigQuery, Recommender Models, Data Engineering

– Data audit (quality, missing values, completeness). – Built ML-based segmentation models to replace classic marketing segments. Technologies: R, Python, RShiny, Salesforce Marketing Cloud, Clustering & Classification Models

– Modèle de détection de fraude avec la base de données graphique (Neo4j) – Built fraud detection models using Neo4j graph algorithms. – Behavioral analysis of beneficiary journeys. – Risk prediction models. Technologies: R, SparkR, Hive, Hadoop, Impala, Neo4j (Cypher queries)

– Behavioral clustering and next-purchase prediction. – Markov Chain modeling for customer transitions. – CLV calculation. Technologies: Python, R, Azure Machine Learning, Logistic Regression, Markov Models