Manuel Caipo, M. Sc.

About me

Manuel Alberto Caipo Ccoa is an external doctoral researcher at the Chair of Sustainable Mobile Drive Systems at the Technical University of Munich (TUM) as part of an industrial PhD project in cooperation with Mercedes-Benz AG in Sindelfingen. His research focuses on the development of multimodal learning methods for fault analysis in software-intensive systems, with applications in electric drivetrain diagnostics.

 

PhD Topic

Context-Aware Multimodal Learning for Fault Analysis in Software-Intensive Systems with Application to Electric Drivetrain Diagnostics
at Mercedes-Benz AG, Sindelfingen, eDrive Software System Integration Department, RD/EDD

 

Research Interests

• Multimodal Machine Learning: Fusion of textual data and time-series signals for fault classification
• Industrial AI & Predictive Maintenance: Anomaly detection, degradation modeling, and Remaining Useful Life (RUL) prediction
• Physics-Aware and Structured Machine Learning: Graph Neural Networks, causal modeling, and Explainable AI
• Scalable ML Systems: Cloud-based data pipelines, MLOps, and Cyber-Physical Systems

Announcements

The following Master's thesis topics are available in cooperation with Mercedes-Benz AG in Sindelfingen, with a planned start date from February 2027 onwards. Interested students are kindly requested to get in touch at least two months before their preferred starting date. For the application, applicants are asked to provide a brief description of their background and motivation. In addition, a current transcript of records may be attached as a reference for the company's application process. Motivation, willingness to learn, and the desire to actively contribute to the research topic are considered particularly important.

 

Topic 1: Multimodal Data Compression for Fault Diagnosis: Text and Time Series Benchmarking

Investigation of compression techniques and optimized embedding strategies for textual and time-series data in the context of fault diagnosis. Evaluation using industrial benchmark datasets, as well as exploration of vision-based approaches for time-series representation and summarization.

 

Topic 2: Graph-Based Intelligent Workflows for Time Series Classification with Textual Enrichment

Development of graph-based methods for knowledge representation and intelligent workflow orchestration in time-series classification. Integration of textual contextual information to improve classification performance in industrial application scenarios.

 

Applications and inquiries should be sent directly to:

manuel_alberto.caipo_ccoa@mercedes-benz.com