A full-stack portfolio project: an AI-driven, vibe-aware café recommender with map UI, busy-hour detection, and optional auto-order agent.
It uses local embeddings for retrieval and Groq (LLaMA-3) for natural language ranking/explanations — a completely free stack for prototyping.
- Vibe-based recommendations (top 3) using semantic retrieval + LLM re-ranking
- Interactive map with Google Maps integration (markers, distances, directions)
- Busy-hour detection (heuristic + optional external popular-times integration)
- Auto-order agent (simulated by default; can be wired to real ordering APIs)
- Local, free embeddings (sentence-transformers) + Groq LLaMA-3 for reasoning