Greedy Routes

This repository contains the Greedy Routes for multi-scenario route optimization using Google’s Route Optimization API and a multi-LLM approach.

Project Structure

greedy-routes/
├── app/
│   ├── __init__.py                 # Makes `app` a Python package
│   ├── main.py                     # FastAPI entrypoint
│   ├── orchestrator.py             # Orchestrates LLM calls & scenario logic
│   ├── router_llm.py              # Router LLM for scenario classification
│   ├── scenario_specialist_fleet.py
│   ├── scenario_specialist_tsp.py
│   ├── scenario_specialist_pickup_delivery.py
│   └── ...                         # Additional scenario files
├── google_route_api.py             # Wrapper to call Google Route Optimization API
├── post_processor.py               # CSV/PDF or other output formatting
├── tests/
│   └── test_router_llm.py          # Example pytest file
├── requirements.txt                # Python dependencies
├── Dockerfile                      # Docker build instructions
├── cloudbuild.yaml                 # CI/CD pipeline for Cloud Build → Cloud Run
└── README.md                       # This file

Quick Start

1. Clone or Pull the repo:

   git clone https://github.com/<YourUsername>/<YourRepoName>.git

2. Install Dependencies locally or in Cloud Shell:

   pip install -r requirements.txt

3. Run FastAPI Locally:

   uvicorn app.main:app --host 0.0.0.0 --port 8080

   Then open your browser at http://127.0.0.1:8080.

4. Test with Pytest:

   pytest

   This runs any tests under tests/.

Deployment (Cloud Run)

• Build & Push via Cloud Build (manual one-off):

  gcloud builds submit --config=cloudbuild.yaml .

• Or set up a trigger in Cloud Build to deploy automatically when pushing to main.

Environment Variables

• GEMINI_API_KEY: PaLM/Gemini key for the Router LLM calls.
• PROJECT_ID: Google Cloud project ID (if you need it in scripts).

For secure secrets, use Google Secret Manager and --set-secrets when deploying to Cloud Run.

TODO

• Implement real calls in google_route_api.py to the Google Cloud Fleet Routing API.
• Add parameter extraction logic in each specialist (Fleet, TSP, Pickup-Delivery, etc.).
• Build a simple React front-end to chat with the /api/optimize endpoint.
• Handle multi-turn user interactions (time window requests, capacity constraints, etc.).

---

Happy coding, and welcome to Greedy Routes!