Amine TELLACHE

– Customer Churn Prediction: Built and deployed an XGBoost classifier to predict churn across multiple products. Collaborated with CSM and Data teams to identify key features; extracted and processed data via Snowflake, dbt, and Airflow. – Forecasting & Analytics: Worked directly on Snowflake-based forecasting pipelines, interpreting and explaining model outputs for business stakeholders. – Automated Incident Response: Designed and deployed an LLM-powered system leveraging OpenAI API and a locally fine-tuned LLaMA model with RAG techniques (LangChain, LangGraph). Implemented a hybrid retrieval mechanism combining NLP similarity search in a CTI vector database with API queries to VirusTotal, MISP, and CrowdStrike. Designed a Chain-of-Thought reasoning architecture to enhance operational workflows. Research paper – Intrusion Detection Systems: Designed a federated multi-agent reinforcement learning architecture to improve accuracy and adaptability of IDS, using PyTorch and the Flower Federated Learning framework. Research paper

– Taught Java and Object-Oriented Programming labs, guiding students in applying OOP concepts through hands-on projects and providing personalized technical support.

– Conducted ITGC (security, change management, operations) and ITAC testing in SOX / CAC / Co-CAC environments. – Audited SAP controls for purchase, sales, and service processes; conducted quality reviews for KPMG UK engagements in collaboration with EMA-region teams. – Assisted in BlackLine ERP implementation and configuration. – Clients included CFAO, Groupe PSA, Knorr-Bremse, L'Oréal, Le Duff, and Buffalo Grill.

– Design and implementation of a Multi-agent DQN (Deep reinforcement learning) using Pytorch, approach for resource allocation in dense LoRaWAN networks implementing network slicing to ensure QoS (Research paper). – Deployed on real hardware (LoRa Shield SX1276, Arduino Uno, Open source LoRaWAN server ChirpStack) with custom Golang code;

– Design and implementation of a mobile application for consulting and synchronising invoices from a web application in real time. The application was developed using React Native for the front-end, REST API with JAVAEE, PostgreSQL.