HackTheFuture – AI-Driven Transit Planner 🚀

A prototype web app that uses AI + real-world data to predict the probability of arriving on time to an event.
It integrates:

• NLP: Understands your text input like "I need to reach my interview by 6:30pm"
• Risk engine: Combines historical transit delays, traffic, weather, and route features to estimate arrival probability.
• Fallback logic: Looks realistic even when historical data is missing.

---

Features

• Enter a natural language plan for reaching a destination.
• AI parses target time and urgency.
• System computes an on-time probability using historical + live data.
• Demo mode included with a pre-set prompt for instant showcase.

---

Prerequisites

• Node.js 18+
• npm
• PostgreSQL database (for historical/traffic/weather data)
• OpenAI API key (for NLP parsing)

---

Setup & Run (localhost)

1️⃣ Install dependencies

# Linux
sudo apt install npm -y

# Mac
brew install npm

### 2️⃣ Configure environment

Create a `.env` file in the root directory with your OpenAI API key and database URL:

```env
CEREBRAS_API_KEY=your_api_key_here
DATABASE_URL=postgres://user:password@host:port/dbname
> If you don’t have real data yet, demo mode uses fallback values so it still works.

## 3️⃣ Run the backend

```bash
node src/index.js

• Server runs on http://localhost:3000
• Demo prompt automatically runs and prints:
  • Parsed intent from AI
  • Estimated probability of arriving on time

API Endpoints

• GET /api/risk?route_id=ROUTE&stop_id=STOP → Returns risk score & delay model
• POST /api/nlp → Parse natural language input into structured intent

Notes for Judges

• Fallback logic ensures realistic outputs even with missing data.
• Fully demoable without real API keys or database.
• Focuses on AI-driven planning + risk estimation.

Quick Demo Tip

To see your AI + risk engine in action instantly:

node src/index.js

Check the console output for parsed intent and arrival probability.