Skip to content

2026 · Build

Data Science & Full-Stack Developer Intern

Designed and shipped an AI web application and its Backend-For-Frontend within the Data Value team, plus a team-wide NLP library for unstructured documents.

The challenge

Within Generali France's Data Value team, I built a tool that extracts actionable insights and recommendations from expertise reports for the company's intermediaries — owning the data model, the VueJS interface, the REST APIs, the Backend-For-Frontend and the document-ingestion pipeline feeding the AI backend.

What I did

  1. Web app architecture: designed, coded and deployed the end-to-end architecture of the web application and its Backend-For-Frontend (BFF).
  2. Full-stack: created the data model and built the frontend in VueJS with REST APIs; the BFF manages user interactions and conversations and routes data to the AI backend (LangGraph).
  3. NLP pipeline: developed a team-wide Python library standardizing the ingestion of unstructured documents — OCR, agentic pre-processing filtering out 50–55% of irrelevant files, semantic chunking, multimodal embeddings, vector database indexing and vector search.
  4. AI agents: collaborated with the team on the development and integration of 4 AI agents automating complex data structuring and analysis tasks.