Overview
Built during the 24-hour innovAIte 2025 hackathon, Pydeia explores how conversational AI can help high-school students navigate university choices in Italy. A React interface guides the conversation, while a Flask backend builds a student profile and ranks courses.
From Conversation to Recommendations
- Listen and respond: browser speech recognition captures answers, and ElevenLabs provides spoken responses.
- Build a profile: structured language-model prompts extract preferences and assign weights to the student's priorities.
- Rank options: embeddings represent academic interests, aspirations, and lifestyle. Similarity scores are combined with practical factors such as budget and location.
- Explain the shortlist: the results view presents three recommendations with scores, key details, and pros and cons.
Prototype Scope
The university dataset is synthetic and intended for demonstration and testing. The project illustrates the recommendation workflow rather than providing a verified catalogue of courses.
Explore the source code, project report, and setup guide.