Sentivault is a production-ready NLP Inference Service designed for real-time sentiment analysis and operational system monitoring. Unlike standard AI scripts, Sentivault features a decoupled architecture that separates the processing engine from the administrative monitoring dashboard.
https://sentivault-nlp-monitor.onrender.com
- Decoupled Architecture: Modular Python design with a clean separation between the NLP logic (
analyzer.py) and the UI layer (app.py). - Live API Simulator: An integrated "Live Tester" to send raw string payloads to the inference engine and receive structured JSON responses.
- Operational Audit Trail: Automated session-based logging using Pandas to track every request with timestamps, payloads, and classification metadata.
- System Health Metrics: Interactive data visualization via Plotly to monitor the distribution of system verdicts and classification accuracy in real-time.
- Low-Latency Processing: Optimized using the VADER engine for near-instant (sub-20ms) response times without external API overhead.
- Inference Engine: Python 3.11, VADER Sentiment
- Dashboard/UI: Streamlit (React-based Python Framework)
- Data Layer: Pandas (Log Management & State Handling)
- Visualization: Plotly Express
- CI/CD & Hosting: GitHub, Render (Automated Deployment)
- Payload Entry: The user/system sends a text payload through the Dashboard or API simulator.
- Logic Execution: The
analyzer.pymodule processes the text using rule-based sentiment reasoning. - Audit Logging: The system captures the request metadata (Timestamp, Latency, Result) into a Pandas DataFrame.
- Visual Monitoring: The Dashboard pulls the stateful log data to update the Health Charts and Audit Table.
| Live API Tester (JSON) | System Audit Logs | Operational Health Chart |
|---|
- Clone the repository: https://github.com/varshinimopidevi/sentivault-nlp-monitor.git Install dependencies: pip install -r requirements.txt Run the Monitor: streamlit run app.py
👤 Author Varshini Mopidevi