Back to Projects
Backend Systems2024
Enterprise Incident Triage System
LLM-powered incident triage with FastAPI. Reduced triage latency.
About This Project
Built a production-grade incident triage system using LLMs and FastAPI for enterprise network diagnostics. The system analyzes log streams, detects anomalies, and reduces triage latency. Containerized microservices with MongoDB and Docker.
Technologies Used
PythonFastAPIMongoDBDockerLLMs
Technical Deep-Dive: Anomaly Detection Pipeline
The system uses a multi-stage anomaly detection pipeline:
- Real-time log ingestion via FastAPI endpoints
- LLM-powered semantic analysis for pattern recognition
- Historical context retrieval from MongoDB for similar incidents
- Automated triage scoring to prioritize critical alerts
This architecture reduced triage latency by enabling faster root cause identification through semantic search over past outages.