CONSILIA

Legal research requires finding relevant material and understanding how it relates to a question. CONSILIA explores how an AI assistant can support that process using Council of Europe treaties and HUDOC case law.

The project was developed as an interdisciplinary semester collaboration between computer science and law. Its purpose is to support research with structured, source-linked analysis, rather than replace professional legal judgment.

I developed the software side of CONSILIA, connecting the user-facing research experience with backend services, data storage and AI retrieval. My teammate Beatrice Eugenia Gervasini contributed legal analysis and curated knowledge.

The application supports contextual follow-up questions and stores users’ chat histories. These features let users continue a research conversation rather than starting again with each question.

Interface: React and Tailwind CSS provide the web application for asking questions and reviewing responses.

Application and data: Node.js / Express, MongoDB and Python / FastAPI form the backend stack supporting the research workflow and stored conversations.

Retrieval-augmented generation: Semantic search uses OpenAI embeddings to retrieve relevant material. Language models work with the retrieved sources and human-curated legal knowledge to produce structured, source-linked analysis.

The emphasis is on keeping supporting material visible to the researcher. Generated analysis still requires review by someone qualified to assess the legal sources and their applicability.

CONSILIA connects full-stack engineering, semantic retrieval and a specialist research domain. Working across computer science and law helped shape both the software and the knowledge it uses.

Created with Beatrice Eugenia Gervasini, who contributed legal research and knowledge curation.

The public website is planned; availability may depend on the project’s launch status.

Explore more of my work in software development, applied AI and interactive experiences.