AI-Driven Incident Detection and Explanation

Log Sentinel is a production-oriented AI system designed to help engineering and operations teams understand system failures faster and more reliably. The platform ingests raw log files, detects anomalies using deterministic feature extraction and grouping logic, and aggregates them into well-defined incidents with clear time boundaries and contextual metadata. Each incident can be explained using a locally hosted large language model, ensuring data privacy and full control over inference behavior. To maintain safety and reliability, all AI-generated outputs are strictly schema-validated, bounded in scope, and backed by deterministic fallbacks when model confidence or compatibility is limited. The system supports optional LoRA adapters for domain-specific fine-tuning while preserving reversibility and architectural isolation. Log Sentinel includes a clean frontend dashboard connected to real backend APIs, comprehensive documentation, and a deployment-ready backend. The project prioritizes clean architecture, operational robustness, and honest documentation of limitations, making it suitable for real-world use, evaluation, and extension.