Build a full-stack web application called TransitAI, an AI-assisted public transportation route finder for Yangon, Myanmar.
Goal
A user enters a starting location and destination. The system should:
Find the nearest suitable bus stops around both locations.
Find possible bus combinations between them.
Calculate the best routes.
Rank the top 3 routes.
Clearly explain exactly which buses to take, where to get off, where to transfer, walking distance, estimated time, and estimated fare.
Tech
React + TypeScript
Tailwind CSS
Supabase database/backend
Leaflet + OpenStreetMap for maps
Modular architecture so the AI/ML system can be added or replaced later.
UI
Create a polished, modern, responsive interface.
Home page:
Start location search/autocomplete
Destination search/autocomplete
Swap locations button
Find Route button
Preferences: Recommended, Fastest, Least Walking, Fewest Transfers, Cheapest
Results page:
Interactive map
Recommended route highlighted
Top 3 alternative routes
Step-by-step journey instructions
Total travel time
Walking distance
Number of buses
Number of transfers
Estimated fare
"Why this route?" explanation
Example:
START → Walk 400 m → Hledan Bus Stop → Bus 21 → Get off at Sule → Walk 300 m → DESTINATION
Database
Create these Supabase tables:
bus_stops
id
name
latitude
longitude
bus_routes
id
route_number
route_name
description
route_stops
id
route_id
stop_id
stop_sequence
travel_data
id
route_id
from_stop_id
to_stop_id
time_of_day
day_of_week
average_travel_minutes
fares
id
route_id
fare
Route Engine
Model the transportation network as a graph:
Bus stops = nodes
Connections between consecutive stops = edges
Edges contain bus route and travel-time information.
When the user searches:
Geocode the start and destination into coordinates.
Find several nearby bus stops rather than only the single closest stop.
Search the transportation graph using Dijkstra's algorithm or A*.
Include walking segments from the origin to the first stop and from the final stop to the destination.
Include transfers between different bus routes.
Calculate travel time, walking distance, transfers and fare.
Return the best 3 valid routes.
Different preferences should change the route ranking.
Important
Do NOT invent real Yangon bus routes or stops and present them as real information. Use clearly labelled DEMO DATA initially.
Create the application so that real transportation data can later be imported through CSV/database records.
Keep route calculation deterministic and separate from the AI layer. The AI must never invent bus numbers, stops, fares or connections.
Build the working MVP now, including the database schema, frontend, route engine, map, location search, demo data, and responsive UI.
Prefer working locally? You need Node.js and npm — install with nvm.
git clone <this-repository-url>
cd <repository-name>
npm i
npm run dev