Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{199843,
        author = {Varsha Negi and Mahika Rastogi and Riya kumari and Ashu kushwaha and Vandana Tripathi},
        title = {Travelara: Travel Planning Web Application Architecture},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {14727-14733},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=199843},
        abstract = {Contemporary AI-assisted travel planning tools suffer from three recurring engineering deficiencies: direct exposure of vendor API credentials in client-side code, unstructured generative outputs that resist programmatic processing, and the absence of a persistent data layer for user accounts and itineraries. This paper presents the design, implementation, and evaluation of Travelara, a full-stack web application that addresses each deficiency through deliberate architectural choices. The system employs a React 19 and TypeScript frontend communicating exclusively with a Django 5.2 REST Framework backend, which proxies all Google Gemini AI calls server-side and enforces JSON schema compliance on every generative response. A stateless JWT authentication mechanism built on a custom Django AbstractBaseUser model with UUID primary keys secures all protected endpoints. SerpApi provides real-time flight and hotel data via Google Flights and Google Hotels engines, with IATA code resolution handled by the airports data library. A normalized relational schema comprising five entities User, Itinerary Plan, Post, Post Like, and Comment supports trip persistence and community interaction. Evaluation across seventeen integration test cases yields a 100% pass rate. Mean response latency is below 100ms for all CRUD operations, approximately 2.2 s for external travel searches, and 8.7 s for full itinerary generation. A structured quality assessment of twenty AI-generated itineraries by four independent evaluators produces a mean overall satisfaction score of 4.2 out of 5.0. The architecture is deployed on Vercel and is offered as a replicable engineering blueprint for production AI web applications.},
        keywords = {travel planning; large language model; Google Gemini; Django REST Framework; React; JWT authentication; SerpApi; itinerary generation; server-side AI proxy; REST API design.},
        month = {April},
        }