System Health AI is a full-stack intelligent diagnostics platform designed to analyze, understand, and predict system behavior using machine learning, generative AI, and autonomous agents.

This project bridges that gap by combining local telemetry analysis, cloud-based intelligence, and AI-driven reasoning to deliver actionable system insights instead of raw numbers.

• GitHub Repository
• Demo Video
• Screenshots Folder
• Hosted Website

1. Web App

• Streamlit (Python)
• chromadb (For RAG)

2. Desktop App

• PyQt5
• psutils for metric collection
• torch , onnxscript, onnxruntime for ML

3. Database

• Supabase Postgresql
• Typescript (Supabase Edge Function)

graph LR
    User --> Login["Login (Google OAuth)"]

    Login --> WebApp["Web App"]
    WebApp --> Reports["Supabase: User Reports"]
    Reports --> Visuals["Text Report & Analytical Visualization"]

    WebApp --> NewChat["New Chat"]
    NewChat --> ChatQuery["Send Query"]
    ChatQuery --> RAG["RAG (ChromaDB)"]
    RAG --> LLM["LLM"]
    LLM --> Response["Response"]
    Response --> StoreChat["Store Chat & Summary"]
    StoreChat --> Supabase[(Supabase)]

    Supabase --> FetchOldChat["Fetch Old Chat"]
    FetchOldChat --> NewChat

    Login --> DesktopApp["Desktop App"]
    DesktopApp --> Monitor["Monitor System"]
    Monitor --> DataCollected["Data Collected"]
    DataCollected --> MLModel["ML Model"]
    MLModel --> CleanedSummary["Cleaned Summary"]
    CleanedSummary --> UploadData["Upload Raw + Cleaned Data + Metadata"]
    UploadData --> Supabase

    Login --> Logout["Log Out"]
			  
						   					  

• Streamlit --> Blazing Fast Web UI with Easy Deployment and Smooth Visualisation of Reports
• PyQt5 --> Minimal Desktop App, Single executable file, with easy execution and Simple Design

• Google OAuth --> Smooth Login and Sign Up
• Loading user's chats and reports --> postgresql type queries via supabase module in python.
• render:
  • onnx Models hosted for Windows Desktop App

• Supabase:
  • Store chats and user details, alongside System Reports
  • Edge function and RLS policies to maintain strict schema of System Reports

• Multiple LLMs like LLama3b, mistral, etc. (via Groq) for specific operations (responding, summarising,etc)
• Models trained with torch based neural networks (torch.nn) , using Adam Optimiser, ReLU Activation Function and Cross Entropy Loss.
• pth models trained by torch converted to onnx for making application lightweight.

• CRUD Based Chat System: Chats are added, deleted and updated smoothly via Supabase and streamlit.
• Collecting System Metrics via psutils: Standard metrics like memory, cpu, temps, disk usage, etc. are appropriately measured
• Functional Vector DB for RAG: Vector DB is suitably chunkized, and stored with metadata to ensure pinpoint and extremly accurate answers with docs fetched being almost always relevant to the query of user

• Continous Integration between chatbot and local app: Chatbot should also analyse reports on own: backend needs to be tweaked a bit more to make this a reality.
• Collecting Binary LOgs and applying more advanced ML models: ML part can be refined more to analyse binary logs as well, which can help catch even highly professional cybersecurity attacks.

• We can host much heavier models with Cloud Computing Services.
• RAG based Vector DB and desktop app can be expanded to more OS beyond the current 3.
• Multiple Authentication modes can be implemented instead of just Google OAuth.

1. System Monitoring and Safety - Malware classification and anomalous behavior of PC based on memory, CPU and temperature data.
2. Smart Assistant For PC Software fixes - Acting as a laser focused, non-hallucinating assistant to resolve unexplained and erratic behavior of user's PCs (For eg freezing of Windows keyboard due to an accidental key shortcut).

If you wish to run the source code or contribute, follow these steps.

# Clone the Web App branch
git clone https://github.com/Once-1296/Web_App_For_System_Health_Metrics_Analyser.git

# Create Virtual Environment
python -m venv venv
# Activate: source venv/bin/activate (Mac/Linux) or .\venv\Scripts\activate (Windows)

# Install Dependencies
pip install -r requirements.txt

# Run the Application
streamlit run app.py

Streamlit manages secrets differently than the desktop app. You must create a file at .streamlit/secrets.toml.

.streamlit/
└── secrets.toml
app.py
Content of .streamlit/secrets.toml:

Ini, TOML
[supabase]
url = "[https://your-project.supabase.co](https://your-project.supabase.co)"
key = "your-anon-key"

[google_auth]
client_id = "your-google-client-id"
client_secret = "your-google-client-secret"

# Clone the Windows branch
git clone https://github.com/MohdHedayati/Local-app---system-health-metric-analyzer.git

# Create Virtual Environment (Verified on Python 3.12.0)
python -m venv venv 
# Activate venv: .\venv\Scripts\activate

# Install Dependencies
pip install -r requirements.txt

# Run the Application
cd PyQt5
python app.py

The application will not start without API credentials. You must create a data folder inside the PyQt5 directory and add the following two JSON files:

Plaintext
PyQt5/
├── app.py
├── data/
│   ├── client_secrets.json
│   └── supabase_secrets.json

Get this from your Google Cloud Console. It should look like this:

{
  "installed": {
    "client_id": "35809.....apps.googleusercontent.com",
    "project_id": "sodium-primer-48...",
    "auth_uri": "[https://accounts.google.com/o/oauth2/auth](https://accounts.google.com/o/oauth2/auth)",
    "token_uri": "[https://oauth2.googleapis.com/token](https://oauth2.googleapis.com/token)",
    "auth_provider_x509_cert_url": "[https://www.googleapis.com/oauth2/v1/certs](https://www.googleapis.com/oauth2/v1/certs)",
    "client_secret": "GOCS...",
    "redirect_uris": ["http://localhost"]
  }
}

Get this from your Supabase Project Settings.

{
    "PROJECT_URL": "[https://bozt....supabase.co](https://bozt....supabase.co)",
    "service_role_key": "eyJhbG....."
}

• Awwab Wadekar: Once-1296
• Mohammed Hedayati: MohdHedayati
• Taha Valiji: tmvalijib24
• Nathan Dsouza: AsparkArcane

• Amal Verma: amal-verma
• Prathamesh Sankhe: PMS61