PakEco

🌍 PakEco AI

Pakistan Environmental Pollution Monitoring & City Comparison

PakEco AI is a hackathon-ready environmental application with two frontends and a shared Python backend:

• Gradio frontend — gradio_app.py
• Streamlit dashboard — streamlit_app.py
• Shared backend — backend.py
• Gemini AI assistant
• Open-Meteo air-quality + weather APIs

Gradio is used as the interactive AI/demo frontend, while Streamlit provides the richer dashboard experience.

Architecture

                 ┌─────────────────────┐
                 │     User / Browser   │
                 └──────────┬──────────┘
                            │
                 ┌──────────┴──────────┐
                 │                     │
          Gradio Frontend       Streamlit Dashboard
        gradio_app.py            streamlit_app.py
                 │                     │
                 └──────────┬──────────┘
                            │
                       backend.py
                            │
             ┌──────────────┼──────────────┐
             │              │              │
        Open-Meteo      Open-Meteo      Gemini API
       Air Quality       Weather          AI

Features

• Pakistani city selector
• US AQI and European AQI
• PM2.5 and PM10
• Temperature, humidity and wind
• Hourly particulate-matter charts
• Multi-city comparison
• AQI ranking chart
• Gemini Environmental Assistant
• Shared backend for both UIs
• Graceful API/AI errors
• No Open-Meteo API key required for the standard MVP request

Data sources

Open-Meteo Air Quality API: https://open-meteo.com/en/docs/air-quality-api

Open-Meteo Weather API: https://open-meteo.com/en/docs

Gemini API: https://ai.google.dev/gemini-api/docs

The air-quality API documents PM2.5, PM10, European AQI and US AQI variables and explains that the forecast is based on CAMS atmospheric-composition forecast data.

Important limitation

The pollution values are model-based and tied to selected coordinates. They should not be described as official ground-monitoring station measurements.

1. Run locally

Python 3.10+ is recommended because current Gradio documentation requires Python 3.10 or higher.

python -m venv .venv

Windows:

.venv\Scripts\activate

macOS/Linux:

source .venv/bin/activate

Install:

pip install -r requirements.txt

2. Add Gemini API key

Get a key from Google AI Studio.

Windows PowerShell:

$env:GEMINI_API_KEY="YOUR_KEY"

macOS/Linux:

export GEMINI_API_KEY="YOUR_KEY"

Never upload the key to GitHub.

3. Run Gradio frontend

python gradio_app.py

Gradio normally opens a local web interface.

4. Run Streamlit dashboard

Open another terminal in the same folder:

streamlit run streamlit_app.py

5. GitHub deployment

Upload these files:

PakEco-AI/
├── backend.py
├── gradio_app.py
├── streamlit_app.py
├── requirements.txt
├── README.md
├── PRD.md
├── DEMO_SCRIPT.md
├── .gitignore
├── .env.example
├── project.json
└── data/
    └── sample_data.csv

Do NOT upload .env, API keys, or secrets.toml.

6. Streamlit deployment

1. Create a GitHub repository.
2. Upload the project.
3. Open Streamlit Community Cloud.
4. Choose your repository.
5. Set the main file to streamlit_app.py.
6. Deploy.
7. Open Settings → Secrets.
8. Add:

GEMINI_API_KEY = "YOUR_REAL_KEY"

1. Save and reboot the app.

7. Gradio deployment options

For the Gradio frontend, the easiest portfolio/hackathon hosting route is a Hugging Face Space using the Gradio SDK.

Upload:

backend.py
gradio_app.py
requirements.txt
README.md

If the platform expects app.py, either rename gradio_app.py to app.py or configure the entry point according to the host's current instructions.

Add GEMINI_API_KEY as a secret/environment variable in the hosting platform.

Hackathon submission

Recommended submission package:

• Project name: PakEco AI
• GitHub URL
• Streamlit demo URL
• Gradio demo URL
• PRD
• 2–3 minute demo video
• Presentation
• Technology list
• Data-source/AI limitations

Responsible AI

PakEco AI:

• identifies its data sources;
• distinguishes model data from station observations;
• does not intentionally generate pollution measurements;
• avoids medical diagnosis;
• explains uncertainty;
• keeps API keys outside source control.

Attribution

Follow the current Open-Meteo attribution requirements when publishing the application or redistributing its data. See the Open-Meteo documentation for current CAMS/Open-Meteo acknowledgement language.