🚀 AI Support Analytics Platform

A small data project to simulate how support tickets can be ingested, analyzed and monitored.

The goal is to build a simple end-to-end pipeline: from raw data to actionable insights.

---

Stack

• Python (Pandas, SQLAlchemy)
• FastAPI
• PostgreSQL (Docker)
• Grafana
• Docker Compose

---

Project Goal

This project simulates a support analytics platform:

• load support tickets into a database
• expose them via an API
• build dashboards to analyze trends
• highlight critical issues
• trigger alerts when things go wrong

It’s designed to be simple but close to real-world use cases.

---

Architecture

CSV → Python ingestion → PostgreSQL → API → Grafana

---

Features

• Ticket ingestion pipeline (CSV → PostgreSQL)

• API endpoints to explore tickets

• Dashboard with:

  • tickets over time
  • distribution by priority / category / sentiment
  • assigned teams

• 🔥 Top critical tickets (actionable view)

• 🚨 Criticality score (global risk indicator)

• Alert when criticality gets too high

---

Criticality Score

A simple scoring logic to identify risky situations:

• +3 → high priority
• +2 → negative sentiment
• +2 → not resolved

This gives a global score that helps prioritize support workload.

---

Run the project

Start services:

docker compose up -d

Generate data:

python app/ingestion/generate_tickets.py

Load data:

python app/ingestion/load_tickets.py

---

Access

• API → http://localhost:8000
• Grafana → http://localhost:3000

---

Notes

This project focuses on:

• structuring a small data pipeline
• building useful dashboards
• thinking in terms of business value (not just data)

---

Possible improvements

• better classification (ML / NLP)
• real-time ingestion
• SLA tracking
• anomaly detection
• orchestration (Airflow)

---

Author

Philippe Kirstetter-Fender