Skip to content

BittenByPython_TravelGenie.md

Latest commit

 

History

History
221 lines (151 loc) · 4.64 KB

BittenByPython_TravelGenie.md

File metadata and controls

221 lines (151 loc) · 4.64 KB

CoC Inheritance 2025
TravelGenie : AI-Powered Smart Travel Planning System

By Bitten By Python
Table of Contents

📝 Description

TravelGenie is an AI-powered smart travel planning system that generates personalized, structured, and optimized travel itineraries using a locally deployed Large Language Model.

It bridges the gap between generic travel suggestions and realistic, budget-aware planning by combining structured dataset filtering with contextual LLM generation. Built with React, FastAPI, and Mistral 7B Instruct running locally with GPU acceleration, TravelGenie delivers city-restricted and logically sequenced day-wise travel plans.

🔗 Links

🤖 Tech-Stack

🏗️ System Architecture

flowchart TD
U["User Preferences\nDestination | Days | Budget | Category"]
FE["React Frontend (Vite)"]
API["Axios API Communication"]
BE["FastAPI Backend"]
FILTER["CSV Dataset Filtering"]
PROMPT["Structured Prompt Construction"]
LLM["Mistral 7B Instruct\n4-bit Quantized"]
GPU["RTX 4060 GPU - CUDA 12.7"]
OUT["Day-wise Structured Itinerary Output"]

U --> FE
FE --> API
API --> BE
BE --> FILTER
FILTER --> PROMPT
PROMPT --> LLM

LLM --- GPU
LLM --> OUT
OUT --> FE
Loading

Front-end

The user interface is built for clarity and interactivity, ensuring seamless itinerary generation.

Framework: React.js (Vite)
Communication: Axios
Storage: Browser localStorage

Key Features

  • Dynamic chat-based input interface
  • Real-time itinerary rendering
  • Editable travel plans
  • Persistent storage of generated itineraries

Back-end

The backend handles dataset filtering, LLM orchestration, and structured output formatting.

Framework: FastAPI
Model Runtime: HuggingFace Transformers + Accelerate

Core Components

  • Dataset filtering engine (CSV-based)
  • Prompt construction logic
  • LLM inference pipeline
  • Structured response formatter

Database & Machine Learning

Data Layer: Structured CSV dataset (India, USA, Iran cities)
Includes: Climate data, pricing ranges, category metadata

AI Infrastructure

  • Mistral 7B Instruct
  • 4-bit quantization
  • NVIDIA RTX 4060 GPU
  • CUDA 12.7

📈 Progress

Fully Implemented Features

  • Personalized Day-wise Itinerary Generation
  • Budget-Aware Filtering
  • City-Restricted Recommendations
  • Climate Summary Integration
  • Local GPU Inference

Partially Implemented Features / Work in Progress

  • Cloud Deployment
  • External API Integrations
  • Multi-city Route Optimization

🔮 Future Scope

  • Real-time travel API integration
  • Multi-city itinerary optimization
  • User authentication and trip storage
  • Scalable LLM deployment
  • Advanced personalization mechanisms

💸 Applications

  1. Personalized Travel Planning
  2. AI Travel Assistant Systems
  3. Academic AI Demonstration

🛠 Project Setup

Clone Repository

git clone https://github.com/Rehan1604/Travel_Genie-Inheritance-.git
cd Travel_Genie-Inheritance-

Backend Setup

cd backend
pip install -r requirements.txt
uvicorn main:app --reload

Frontend Setup

cd frontend
npm install
npm run dev

Frontend runs via npm run dev.
Backend runs via uvicorn main:app --reload.

👨‍💻 Team Members

👨‍🏫 Mentors