Master Thesis: AI ML for Anomaly Detection in Baseband System Hardware

Om jobbet

Join our Team

About this opportunity:

We are seeking a highly motivated Master's student for a thesis project focused on exploring and analyzing historical component temperature and power consumption measurements from the hardware of our baseband systems. The aim is to explore advanced anomaly detection methods to improve monitoring of baseband systems. By combining time-series analysis, unsupervised/semi-supervised learning, and domain-driven rules, the goal is to design an approach that can reliably detect anomalies and reduce false alarms in real deployments.

What you will do:

• Analyze historical data from temperature and power consumption measurements.
• Investigate methods for unsupervised anomaly detection (HMM, Isolation Forest, Autoencoders, forecasting residuals, etc.)
• Explore weak supervision and synthetic anomaly injection for model validation.
• Evaluate and compare approaches based on accuracy, false alarm rates and detection delays.
• Develop a prototype pipeline that combines statistical rules with ML-based detection

The skills you bring:

• Background in data science, machine learning, and AI methodologies.

• Proficiency in time series analysis and related algorithms.

• Experience with Python and relevant libraries (e.g., pandas, NumPy, scikit-learn, TensorFlow, PyTorch).

• Experience with Apache Spark (pyspark), Jupyter, Linux, SQL(PostgreSQL).

• Familiarity with data visualization techniques and tools, such as Grafana.

• Basic understanding of clustering techniques and statistical methods.

• Knowledge of power consumption and temperature measurement data from electronic hardware products is a plus.

• Ability to work independently and systematically, with a problem-solving mindset.