ResQAI-ML-Models-AI-APIs 🤖

Overview 🌟

This repository is part of the ResQAI system, a disaster response and management platform designed to improve coordination among stakeholders during natural calamities. The ResQAI-ML-Models-AI-APIs repo focuses on AI and machine learning components, offering pre-trained models, Flask APIs, and generative AI integrations to enhance disaster management capabilities.

Check out the main ResQAI repository here: ResQAI GitHub Repo

---

Tech Stack 🛠️

• Programming Language: Python 🐍
• Frameworks: Flask 🌐
• Cloud Platform: Google Cloud Platform (GCP) ☁️
  • Cloud Run for deployment 🚀
  • Google AI Studio 🤖
  • Vertex AI (including AutoML) 📈
  • Gemini 1.5 models (Flash and Pro versions) 🌟
  • Google Search Retrieval 🔍
• Libraries:
  • Scikit-learn 📚
  • Joblib ⚙️
  • Pandas 🐼
  • Matplotlib 📊
• Machine Learning Algorithm: Decision Tree Classifier 🌳

---

Repository Structure 📂

├─── 📄 .gitignore
├─── 📂 Google-GenAI-APIs
│   ├─── 📂 Landing
│   │   ├─── 📄 main.py
│   │   └─── 📄 requirements.txt
│   ├─── 📂 VertexAPI
│   │   ├─── 📄 main.py
│   │   └─── 📄 requirements.txt
│   ├─── 📂 VertexAPIPro
│   │   ├─── 📄 main.py
│   │   └─── 📄 requirements.txt
│   └─── 📂 sample-genai-templates
│       ├─── 📄 Gemini_normal_chatbot_Template.py
│       ├─── 📄 GoogleAIStudio.py
│       └─── 📄 vertexGenaiApi.py
├─── 📂 ML-Models-APIs
│   ├─── 📂 ML-JupyterNotebooks
│   │   └─── 📂 Flood Prediction
│   │       ├─── 📄 2022monthly.csv
│   │       ├─── 📄 Flood Prediction Experiment.ipynb
│   │       ├─── 📄 test.csv
│   │       └─── 📄 train.csv
│   ├─── 📂 ML-Prediction-APIs
│   │   ├─── 📂 Earthquake
│   │   │   ├─── 📄 main.py
│   │   │   └─── 📄 requirements.txt
│   │   └─── 📂 Flood
│   │       ├─── 📄 main.py
│   │       ├─── 📂 models
│   │       │   ├─── 📄 decision_tree_model.joblib
│   │       │   └─── 📄 subdiv_encoder.joblib
│   │       └─── 📄 requirements.txt
│   └─── 📂 ML-model-vertexai-api
│       └─── 📄 main.py
└─── 📄 README.md

---

Folder Breakdown 📁

Google-GenAI-APIs ✨

1. Landing:
   • Contains a Flask API for ResQAI’s chat assistant.
   • Provides real-time updates and disaster management assistance using Vertex AI Gemini 1.5 Flash 001 model.
2. Sample GenAI Templates:
   • Gemini_normal_chatbot_Template.py: Demonstrates a basic chatbot using Gemini.
   • GoogleAIStudio.py: Showcases Google AI Studio API usage in Flask.
   • vertexGenaiApi.py: Provides a Vertex Generative AI API template.
3. VertexAPI:
   • Implements a Flask API using Gemini 1.5 Flash 001 model for general AI tasks like case study agents and AI interpretations.
4. VertexAPIPro:
   • Similar to VertexAPI but uses the Gemini 1.5 Pro 002 model for tasks requiring precise AI outputs like JSON generation, document summarization, and disaster analysis.

ML-Models-APIs 🧠

1. ML-JupyterNotebooks:
   • Contains datasets and a Jupyter notebook (Flood Prediction Experiment.ipynb) to analyze various algorithms for flood prediction.
   • Selected model: Decision Tree Classifier, exported using Joblib.
2. ML-model-vertexai-api:
   • Flask API template for deploying ML models using Vertex AI endpoints.
3. ML-Prediction-APIs:
   • Earthquake: Flask API using an AutoML-generated Vertex AI model for earthquake predictions.
   • Flood: Flask API leveraging the exported Decision Tree Classifier for flood predictions, integrating Google Search Retrieval for human-readable outputs.

---

Installation 🛠️

Note 📌

All folders containing APIs (e.g., main.py and requirements.txt) follow a similar installation process. This includes creating a virtual environment, installing dependencies, configuring the .env file, and running main.py. Refer to the respective folder for additional context.

Prerequisites 📋

• Python installed 🐍.
• Google Cloud service account with necessary permissions 🔐.

Steps 🔧

1. Navigate to the desired folder (e.g., cd Google-GenAI-APIs/Landing).
2. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate   # For Linux/Mac
   venv\Scripts\activate    # For Windows

3. Install dependencies:

   pip install -r requirements.txt

4. Configure environment variables in a .env file:

   GOOGLE_GENAI_API_KEY=<Your Google AI API Key>
   GOOGLE_CREDENTIAL_FILE=<Path to your GCP JSON key>
   PROJECT_ID=<Your Google Cloud Project ID>
   ENDPOINT_ID=  # For Earthquake prediction API

5. Run the API:

   python main.py

6. Access the API locally at http://127.0.0.1:5000.

---

Deployment 🚀

All APIs are deployed on Google Cloud Run ☁️, ensuring scalability and reliable performance during disaster management operations.

---

License 📜

This project is licensed under the MIT License 📄.

---

Contribution 🤝

Feel free to fork the repository, raise issues, and submit pull requests for enhancements or bug fixes 🛠️.