MobAi 2025

MobAi is a book-focused application repository with three main parts:

• Ai/final: a FastAPI service that powers book search, recommendation, and evaluation workflows.
• FrontEnd/app: a Flutter application that consumes the product and AI services.
• BackEnd: backend notes and supporting documentation for server-side work.

The repository is organized around a book catalog experience, combining semantic search, recommendation logic, and mobile UI work.

Repository Structure

MobAi/
├── Ai/
│   └── final/
│       ├── app.py
│       ├── search.py
│       ├── recommend.py
│       ├── evaluator.py
│       ├── books.json
│       ├── test_results.json
│       └── .env.example
├── BackEnd/
│   └── ReadMe.md
├── FrontEnd/
│   └── app/
│       ├── lib/
│       ├── android/
│       ├── ios/
│       ├── web/
│       └── windows/
└── README.md

Main Components

AI Service

The AI service is a FastAPI app that exposes two endpoints:

• POST /search: runs the book search pipeline.
• POST /recommend: returns recommendations based on a user profile and order history.

The search pipeline uses:

• SentenceTransformer embeddings for semantic matching.
• FAISS for similarity search over the book catalog.
• LangChain and Gemini models for prompt classification and result filtering.

Flutter Frontend

The Flutter app lives under FrontEnd/app and contains the main user interface, app state, routes, models, and widgets.

Evaluation Script

The AI folder also includes an evaluator that generates test prompts, runs them through the search chain, and writes structured results to test_results.json.