ChatBot REX
Lessons-learned records scattered across spreadsheets and email. A chatbot and a KPI dashboard now centralize them for engineers, inside the internal network.
ChatBotUV
Rich cultural content locked in print magazines and PDFs. A chatbot answers concrete questions using only that sourced, published content.
- Role
- Design and development
- Period
- Summer 2026
- Setting
- Personal project for Popamine
- Python
- PyMuPDF
- Sentence-Transformers
- FastAPI
- Groq (llama-3.3-70b)
- React
- TypeScript
- Vite
- Tesseract OCR
In action
Context
Popamine, publisher of the UV magazines (city guides) and Le Haut Parleur, holds rich cultural content about Saint-Nazaire and the Loire-Atlantique region — but scattered across print and PDF material that is hard to query beyond linear reading.
Mission
Built a complete RAG pipeline: content extraction (PDF scraping for Le Haut Parleur, OCR via Tesseract for scanned UV magazines), chunking, embedding indexing computed locally (CPU, Sentence-Transformers), cosine similarity search, and grounded, sourced answer generation via Groq (llama-3.3-70b-versatile). React/TypeScript/Vite interface, FastAPI backend.
Impact
A chatbot able to answer concrete cultural questions (e.g. outing recommendations in Saint-Nazaire) by relying solely on real, sourced published content rather than a language model’s generic knowledge.
Let's talk about your project.
Describe what you need in a few lines; we'll see together whether I can help.
Get in touch