AI-Powered Log Monitoring and Automated Remediation using OpenTelemetry

Distributed microservice architectures produce massive volumes of logs that make manual monitoring inefficient and time-consuming. Conventional monitoring tools rely on rulebased alerts and manual inspection, which often fail to detect emerging failures in complex systems. This work presents an intelligent log monitoring framework that combines real-time telemetry collection, machine learning-based anomaly detection, and large language modelbased analysis to improve system observability. Logs are collected using OpenTelemetry and processed through a streaming pipeline built with Apache Kafka and Apache Flink. An ensemble of machine learning models, including Isolation Forest, LSTM, and Prophet, is applied to identify abnormal patterns in log streams. When anomalies are detected, a language model generates interpretable insights and recommended remediation actions. The system stores processed data in Elasticsearch and visualizes insights through Kibana dashboards. Experimental evaluation demonstrates high detection accuracy and low processing latency, enabling faster incident diagnosis and reduced operational overhead.