The aim of this work is to improve defect detection in RACOM's manufacturing pipeline by applying data mining and machine learning techniques to Linux system logs. Currently, the company relies on regex-based keyword matching, which is insufficient for detecting novel errors. To address this, the student will explore and analyze production logs to infer rules and identify complex anomalies. Based on these findings, an automated monitoring solution will be proposed. The selected anomaly detection model will be deployed and tested in RACOM's live production environment to validate its practical effectiveness. The result of the thesis will be a technical report and an implementation of a system for automated anomaly detection. Only selected parts of the program implemented by the student will be available in the IS, exclusively in the form of read-only source code under a proprietary license.