Francesco Paolo De Gregorio Email and Phone Number

CarPlanner is a digital car-as-a-service marketplace for all-inclusive, hassle-free car subscriptions.Launched in 2015 as a MarTech Startup and lead generation partner to leading car manufacturers such as BMW, Stellantis, Mercedes, Renault and Toyota. As the auto industry is shifting from car ownership to car-as-a-service, CarPlanner is evolving as well to become a leading car subscription marketplace.Main activities:- I'm developing an advanced RAG… Show more CarPlanner is a digital car-as-a-service marketplace for all-inclusive, hassle-free car subscriptions.Launched in 2015 as a MarTech Startup and lead generation partner to leading car manufacturers such as BMW, Stellantis, Mercedes, Renault and Toyota. As the auto industry is shifting from car ownership to car-as-a-service, CarPlanner is evolving as well to become a leading car subscription marketplace.Main activities:- I'm developing an advanced RAG (Retrieval-Augmented Generation) chatbot that integrates open source and private LLMs, leveraging reranking (open source model: ms-marco-MiniLM-L-12-v2), prompt engineering, query refinement, and rewriting to deliver precise, structured answers in markdown. It uses techniques like Chain of Thought (CoT) reasoning and few-shot prompting to minimize hallucinations, with a metadata-enhanced vector database for improved search relevance. The bot features a custom web crawler for training on websites and documents, integrates LangFuse (self hosted using docker container) for optimal monitoring, and includes a CMS (Next.js) that allows users to manage and personalize their bot settings. Currently evolving into a SaaS product for diverse industries, it can be embedded on any website. The backend is built with Python (Flask), the frontend with React, and the server API with PHP- Prototyping and development of a Customer Data Platform on AWS. Show less

Projects: - Scoping and development of Generative AI applications. - Fine tuning of LLMs for specialized use cases. - Development of an LLM powered web application using Langchain, Chainlit, Python, Docker (for containerization) and AWS (for Cloud deployment). The application enables users to chat with their documents and obtain detailed summaries. - LLM powered web application that enables users to translate code written in a specific language (e.g. Java) and… Show more Projects: - Scoping and development of Generative AI applications. - Fine tuning of LLMs for specialized use cases. - Development of an LLM powered web application using Langchain, Chainlit, Python, Docker (for containerization) and AWS (for Cloud deployment). The application enables users to chat with their documents and obtain detailed summaries. - LLM powered web application that enables users to translate code written in a specific language (e.g. Java) and translate it into another language (e.g. Python). - Generative AI Learning Coach: LLM powered streamlit application specialized for the teaching use case. The app enables users (employees of an Italian Bank) to quickly master lengthy and complex concepts via simple queries. Show less

Main projects: - Development of ML models for fraud identification in the betting industry (23% reduction of false positives) - ML-based forecasting analysis for a client operating in the consumer electronics sector.- Development of a reporting solution for a multinational corporation operating in the insurance sector - Development of a Python reporting solution for a multinational corporation operating in the insurance sector (using Python, Sql, NoSql, data science and… Show more Main projects: - Development of ML models for fraud identification in the betting industry (23% reduction of false positives) - ML-based forecasting analysis for a client operating in the consumer electronics sector.- Development of a reporting solution for a multinational corporation operating in the insurance sector - Development of a Python reporting solution for a multinational corporation operating in the insurance sector (using Python, Sql, NoSql, data science and visualization libraries). - Customers Segmentation based on behaviors & interactions Show less

Worked in the Data and Analytics department.Main activities:- Worked on data analysis and time series forecasting for a client operating in the consumer electronics industry- Used Databricks, Python (and related Data Science stack) and time series forecasting libraries such as statsmodels and Prophet