An end-to-end AI-powered auditing pipeline developed for the Accenture Databricks Hackathon. This system processes large-scale medical facility data from the Virtue Foundation (Ghana) using Llama 3.1 via Groq to automatically extract clinical capabilities and flag infrastructure risks.
# 🤖 VirtueAudit: AI-Automated Medical Infrastructure Verification & Risk Assessment
> **An end-to-end AI-powered auditing pipeline for automated medical infrastructure verification and risk assessment.**
> *Built for the Accenture × Databricks Hackathon*
---
## 📖 Overview
**VirtueAudit** automates the verification and risk assessment of medical infrastructure by processing large-scale facility data from the **Virtue Foundation (Ghana)**. It leverages **Llama 3.1 via Groq** for sophisticated natural language understanding and **Databricks** for scalable data processing — enabling automatic extraction of clinical capabilities and proactive flagging of infrastructure risks.
The project runs in two modes:
- 🔴 **Full Apache Spark pipeline** on Databricks for production scale
- 🟢 **Pandas-based local fallback** for development
Both modes feed the same interactive **Streamlit dashboard**, deployed on Databricks Apps and Streamlit Cloud.
---
## ✨ Features
| Feature | Description |
|---|---|
| 🤖 **AI Risk Assessment** | Llama 3.1 via Groq analyzes descriptions and flags infrastructure risks |
| 🏥 **Capability Extraction** | Auto-extracts and categorizes clinical services per facility |
| ⚡ **Dual Execution Engine** | Apache Spark on Databricks + automatic Pandas fallback |
| 📊 **Interactive Dashboard** | Streamlit app with KPIs, Plotly charts, and per-facility case views |
| 📄 **Structured Audit Reports** | Outputs `Virtue_Final_Audit.csv` with 6 fields including `is_suspicious` |
| 🐳 **Containerized Dev** | `.devcontainer` for consistent, one-click developer setup |
---
## 🌐 Live Deployments
| Platform | URL | Engine |
|:---:|:---:|:---:|
| 🔴 **Databricks Apps** | virtue-audit-dashboard-...databricksapps.com | Apache Spark |
| 🟢 **Streamlit Cloud** | ghana-medical-audit.streamlit.app | Pandas |
---
## 🛠️ Tech Stack
**Core AI / ML**
**Dashboard**
**Cloud Infrastructure**
**Dev Environment**
---
## 📂 Project …