TRITON INSIGHTS 2.0

A Streamlit-based analytics dashboard for querying, visualizing, and analyzing environmental sensor and alert data from AWS Athena and Bedrock LLM. Supports both LLM-based natural language queries and manual SQL filter selection.

---

Features

• Natural Language Data Querying:
  Query environmental and alert data using Bedrock LLM (Cohere Command-Light) for Athena SQL generation and natural language answers.

• Manual Athena SQL Interface:
  Direct selection of device, columns, and date range with visualization and table export.

• AWS Integration:

  • Athena for query execution
  • S3 for result storage
  • Glue Data Catalog
  • (Optionally) Bedrock Vector Knowledge Base

• Interactive Visualizations:
  Time series and multi-metric plots with Plotly in Streamlit.

---

File Structure

• main_app.py   # Streamlit entrypoint (mode selector, logo, routing)
• llm_ui.py     # UI for LLM-driven (natural language) queries
• nonllm_ui.py   # UI for manual queries
• athena_utils.py  # Athena SQL runner, result handling, Plotly plotting
• llm_utils.py    # Bedrock LLM wrapper and prompt logic
• knowledge_base.py # Mappings and example queries for both tables
• image.png     # App logo

---

Requirements

• Python 3.9+
• AWS credentials/configured for Athena, Bedrock, S3, Glue
• The following Python packages (see requirements.txt):

streamlit
boto3
pandas
plotly
Pillow