DF Bus — Real-Time Mobility Platform

Public engineering showcase of a production application. The production backend and frontend repositories are private by design.

Overview

DF Bus is a real-time public transportation platform focused on Brasília, Brazil.

The product combines mobility data, geospatial processing, real-time vehicle telemetry, route and stop discovery, progressive web technologies, and conversational AI in a single user-facing application.

The goal is to make public transport information easier to explore while also demonstrating how backend engineering, real-time systems, mapping, product development, and AI integration can work together in production.

Product Capabilities

• Search public transportation routes
• View route paths on an interactive map
• Track vehicles in near real time
• Find nearby stops
• Explore routes serving a selected stop
• View timetables
• Save favorite routes
• Filter vehicles by direction
• Install and use the application as a PWA
• Query transportation information using natural language

High-Level Architecture

flowchart LR
    A[Mobility Data Sources] --> B[Normalization & Processing]
    B --> C[FastAPI Backend]

    C --> D[Routes]
    C --> E[Stops]
    C --> F[Vehicle Telemetry]
    C --> G[Timetables]

    D --> H[Web / PWA Client]
    E --> H
    F --> H
    G --> H

    H --> I[Interactive Map]

    J[Conversational AI] --> K[Tool / Function Calls]
    K --> C
    C --> J

Loading

Backend Engineering

The backend exposes application capabilities through REST APIs and is responsible for processing and serving structured mobility data.

Key responsibilities include:

• route discovery
• stop information
• timetable processing
• vehicle telemetry
• geospatial queries
• data normalization
• external data integration
• caching and resilience strategies
• structured logging

Main Technologies

• Python
• FastAPI
• REST APIs
• SQL
• Async I/O
• Geospatial processing
• Cloud deployment
• Structured logging

Frontend Engineering

The client is designed around a map-first experience and mobile use.

Key capabilities include:

• interactive maps
• route polylines
• real-time vehicle markers
• marker clustering
• responsive UI
• light and dark themes
• favorites
• route and stop interaction
• PWA installation
• continuous vehicle updates

Main Technologies

• JavaScript
• Leaflet
• MarkerCluster
• Progressive Web App
• HTML/CSS
• REST integration

Conversational AI

DF Bus also explores natural-language interaction with transportation data.

Instead of depending only on free-form LLM responses, the AI can invoke application functions to retrieve structured route, stop, timetable, and journey data.

User question
      ↓
Intent interpretation
      ↓
Function / Tool Call
      ↓
DF Bus API
      ↓
Structured mobility data
      ↓
Natural-language response

This approach lets the model operate as an interface to application data rather than as an independent source of transportation facts.

Real-Time Engineering Challenges

Vehicle positions and upstream transport data change continuously, so the application needs to handle real-world uncertainty rather than only static datasets.

Engineering concerns include:

• frequent data refreshes
• stale vehicle positions
• map marker synchronization
• network instability
• upstream API inconsistencies
• client-side performance
• mobile bandwidth and battery constraints
• graceful degradation when a data source is unavailable

Product Engineering

DF Bus is not only a technical demonstration. It has been deployed as a real application and used by public transportation users.

The project brings together:

Backend Engineering + Real-Time Systems + Geospatial Data + Frontend Development + Product Engineering + AI Integration

What Is Intentionally Not Public

To protect the production system and proprietary implementation, this repository does not include:

• production backend source code
• production frontend source code
• infrastructure configuration
• API credentials
• deployment secrets
• proprietary processing logic
• internal telemetry
• private production datasets
• security-sensitive operational details

This repository exists to demonstrate product, architecture, and engineering decisions without publishing the implementation itself.

My Role

I designed and developed the product across backend, frontend, integrations, real-time data flows, and conversational AI capabilities.

Portfolio Context

This repository is a showcase of the production system.

The actual backend and frontend codebases remain private intentionally. The public material focuses on architecture, engineering challenges, product capabilities, and technical decisions while preserving the implementation and intellectual property.

---

Production source code is intentionally private. Public documentation provided for portfolio purposes only.