ou4y/itsm-ai-service repository overview

This image provides a production-ready AI microservice designed for IT Service Management (ITSM) platforms. The service predicts the probability of Service Level Agreement (SLA) breaches before they occur and exposes REST APIs for real-time inference, administrative insights, and feedback-driven model improvement.

The service is built as a standalone microservice and is intended to run as part of a containerized system using Docker. It integrates with external infrastructure services such as MySQL for persistent storage and RabbitMQ for event-driven retraining workflows.

Key Features: • SLA Breach Risk Prediction Exposes REST endpoints to predict the likelihood of SLA violations for incoming tickets using structured ticket metadata.

• Admin Feedback Collection Supports human-in-the-loop workflows by storing administrator decisions and final outcomes for continuous learning.

• Event-Driven Retraining Uses RabbitMQ to asynchronously trigger model retraining when sufficient new feedback is available, without blocking API requests.

• Model Retraining Pipeline Includes offline batch retraining logic with class-imbalance handling and safe model replacement.

• REST API with OpenAPI Documentation Built with FastAPI and provides interactive API documentation via Swagger UI.

• Containerized and Portable Fully containerized using Docker, enabling consistent behavior across development and production environments.

Running the Service 1- Quick Start (Standalone) docker run -p 8000:8000 /itsm-ai-service

2- API docs: http://localhost:8000/docs⁠

Recommended Setup (Docker Compose) 1- docker compose up --build The AI service communicates with dependent services using internal Docker networking.

API Access

Once running, the service exposes: • Prediction endpoints for SLA breach probability • Feedback endpoints for administrative decisions • Dashboard endpoints for risk insights