Design and Development of AI Agent-Based Travel Planner Platform using Agile Scrum and DevOps Integration
Rahman, Abdur
Abstract
Planning a trip often turns into a tedious chore as travelers are forced to jump between disjointed platforms to organize itineraries, evaluate accommodations, and execute bookings. This project resolves that complexity by delivering a cloud-native, end-to-end travel platform powered by autonomous AI agents. Unlike traditional engines that merely provide static recommendation lists, this system uses intelligent agents to interpret individual user preferences, dynamically generate personalized travel itineraries, and autonomously execute complex booking workflows. The frontend is built with Next.js to ensure a responsive, seamless user experience, while Supabase serves as the backend infrastructure, providing flexible database management and vector search capabilities to power high-dimensional context retrievals for the AI models. To handle the operational demands of a multi-service architecture, the project was managed using the Agile Scrum framework, enabling continuous iterative development and feature delivery. Robust DevOps practices were integrated from the ground up to establish an automated CI/CD pipeline using Jenkins, streamline application containerization via Docker, and orchestrate scalable container deployments across Amazon Elastic Kubernetes Service (AWS EKS). Managed via Infrastructure as Code, this automated pipeline guarantees consistent testing, rapid deployment, and high system availability under varying workloads. Ultimately, this capstone project demonstrates how synthesizing agentic AI architectures with modern cloud-native engineering principles can transform a fragmented, overwhelming workflow into a cohesive, intelligent, and effortless travel planning solution.
Permanent link — send this address to others:https://lib.su.edu.bd/thesis/123456789/3069