Giuseppe Paolini

AI/ML Engineer in Mango Data Science Dept. Designed, implemented, and validated LLM-based multi-agent systems without ground-truth datasets, leveraging self-reflection and self-consistency techniques to improve reliability and autonomous reasoning. Led AI & Data Reporting: an AI-driven batch analytics project using Apache Spark (Python) to generate automated KPI reports delivered via Streamlit dashboards for multiple business stakeholders, enabling timely identification of trends to support data-driven decisions.

LLM CoE - R&D Dept: Design and development of generative artificial intelligence software (genAI), integrating multimodal Large Language Models (LLMs). Data Science: Conducted exploratory data analysis (EDA) on various datasets to identify critical patterns and correlations; applied statistical techniques and machine learning models for predictive maintenance, enabling proactive prevention of equipment failures. AI Agent Development: Implemented autonomous AI agent systems with contextual understanding and adaptive learning, including single and multi-agent architectures for process automation and Retrieval-Augmented Generation (RAG) using libraries such as SmolAgents, LlamaIndex, LangGraph, crewAI, and Autogen. Executed a dual-agent system for supply chain management that retrieves data from multiple databases, performs analytics, provides comprehensive responses, and maintains conversational memory to improve user interaction over time.

Master’s Thesis project: Development of machine learning models and frameworks ex-novo for brain tumor detection in MRI images.