Travel Itinerary

• 

  Travel Itinerary

• 

  branch-and-bound-tree

• 

  local-guide-form

• 

  osrm-distance-table

• 

  osrm-network-graph

• 

  planner-chat

• 

  route-flow-map

The idea for Travel Itinerary came from a common problem travelers face. Many tourists, especially foreign visitors, don’t know the exact names of places to visit or how to plan an efficient route. I wanted to build a platform that could guide travelers based on their interests and automatically plan their journey.

This hackathon was a great learning opportunity for me. While building this project, I researched how popular map platforms handle route planning. I explored the Travelling Salesman Problem and different algorithms used to find the most efficient route, and finally implemented the Branch and Bound algorithm.

Another interesting part of the project was integrating AWS Nova models through Nova Act to power the AI travel assistant. The Nova model understands user queries written in natural language, allowing travelers to interact with the system as if they were chatting with a real person.

Instead of training a model, I implemented a RAG based architecture where relevant travel information is retrieved from a vector database and structured place data. This allows the system to use real time travel information when generating recommendations.

I also used Leaflet.js to visualize routes on the map. Overall, the project became a combination of AI assistance, route optimization algorithms, retrieval based knowledge systems, and interactive mapping technologies.

Built With

• astravectordb
• gemini-embedding
• javascript
• leaflet.js
• nextjs
• novaact
• novaai
• openaisdk
• postgresql
• prisma
• rag