AI-Powered Tourism Analytics

Overview

Built as a production-grade ML system with a decoupled microservices architecture, this project combines computer vision and recommendation workflows. The repository README describes EfficientNet-based heritage classification, FAISS-backed recommendations, MLflow observability, and quality-gated training pipelines with FastAPI and Next.js services.

Key Highlights

• EfficientNetB0 transfer-learning classifier for historical structures.
• Item-based collaborative filtering with FAISS vector search for recommendations.
• MLflow-based experiment tracking and model serving workflow.
• Data quality gates to prevent training on corrupted datasets.
• Decoupled FastAPI + model-server + Next.js architecture for independent scaling.

Technology Stack