Req ID:524034 

 

At Alstom, we understand transport networks and what moves people. From high-speed trains, metros, monorails, and trams, to turnkey systems, services, infrastructure, signalling and digital mobility, we offer our diverse customers the broadest portfolio in the industry. Every day, more than 80 000 colleagues lead the way to greener and smarter mobility worldwide, connecting cities as we reduce carbon and replace cars.

 

 

We are offering a Master’s thesis opportunity in collaboration with academic and industrial partners linked to the MONA LISA research initiative. The thesis focuses on how an AI agent can help engineers interpret log data, identify relevant patterns, connect signals to system context, and generate clear technical explanations and visual summaries that support issue understanding.

The topic is closely aligned with one of the documented Alstom / MONA LISA use cases, which explicitly describes AI-supported visualisation of logs as an important step towards faster understanding of reported product issues. It is intended as a Master’s-thesis project for students interested in LLMs, agent development, industrial analytics, multimodal AI, and visual decision support.

 

Background

The MONA LISA initiative addresses the challenge that cyber-physical-system development is often fragmented across hardware, software, models, logs, documents, and lifecycle phases. One of its core ambitions is to use AI-supported semantic analysis and visual analytics to make complex engineering information easier to access and interpret. Within this broader context, the Alstom use case defines a concrete industrial problem: product issues are reported through tickets with associated logs, and engineers need faster, better-supported ways to understand the logs and their relevance to the issue. 

The use-case documentation describes UC-02 Visualise logs, where the AI agent retrieves ticket descriptions and attached log files, analyses them, and generates visualisations intended to improve engineers’ understanding of the issue context. The same document also positions this capability as a precursor to later steps such as root cause analysis and solution proposal development. This makes log interpretation and visual explanation a highly relevant and manageable thesis topic with direct industrial value.

 

Connection to previous thesis work

This thesis also builds directly on earlier student work already completed within the same collaboration. A previous thesis on multi-modal document context search with large language models investigated how AI can retrieve relevant information from engineering documents spanning text, images, tables, and diagrams. That work showed how industrial knowledge is often spread across heterogeneous artefacts and highlighted the importance of multimodal pipelines, retrieval strategies, and domain-sensitive handling of engineering terminology.

The presentation material for that work additionally mentions Dataiku, LangChain, and future interest in agentic search, which signals a clear path from document retrieval towards more active AI-agent assistance. The current vacancy follows that path by moving from “finding relevant information” to “interpreting complex technical evidence and explaining it usefully to engineers”.

The thesis also complements the more recent work on LLM-based multi-agent systems for engineering workflows. That work demonstrated that agent-based architectures can be implemented and evaluated in an engineering context using structured assessment criteria such as completeness, correctness, and traceability. The present topic narrows the focus to one especially promising problem: how an AI assistant or lightweight agent architecture can make raw logs and log-derived evidence easier to understand through explanation and visual support.

 

Thesis objective

The objective of this thesis is to design, implement, and evaluate an AI-based log-interpretation agent for industrial engineering scenarios. The system should help turn low-level log evidence into higher-level technical understanding by identifying patterns, anomalies, or event sequences and relating them to issue descriptions or system context. Depending on scope, the thesis may focus on a single intelligent assistant or on a small agent-based design where one component analyses logs, another retrieves supporting context, and another generates an explanation or visual summary.

The aim is not merely to plot data. The deeper goal is to investigate how AI can support engineers in understanding what matters in the logs, how those findings can be explained in a technically meaningful way, and what forms of explanation are useful in practice. This could include textual explanations, guided summaries, anomaly highlighting, event grouping, or visual representations that connect log patterns to issue narratives.

 

Examples of research questions

Possible research questions include the following.

• How can an AI agent identify and explain relevant patterns in industrial log data associated with reported engineering issues?
• How can log findings be linked to surrounding technical context such as issue descriptions, documentation, or previous incidents?
• What kinds of explanation or visualisation support engineers most effectively when they are trying to understand a complex operational problem?

Should the solution be built as a single assistant or as a small agent architecture with separate components for retrieval, interpretation, and explanation?

How should a system of this kind be evaluated in terms of technical relevance, clarity, faithfulness to the underlying evidence, and usefulness for expert users?

 

Possible scope and work packages

Depending on the student’s interest and available project access, the thesis may include several of the following activities.

• Literature review on AI for log analysis, explanation generation, and visual analytics in engineering contexts.
• Analysis of log-visualisation and issue-understanding requirements from the documented MONA LISA / Alstom use case.
• Design of a log-interpretation assistant or small agent architecture.
• Development of a prototype capable of transforming raw or semi-structured log information into technical summaries or visual explanations.
• Evaluation using example cases or representative scenarios, potentially considering criteria such as clarity, correctness, relevance, and traceability.
• Reflection on deployment constraints, explainability, and how the capability could later feed into root cause analysis and solution-support workflows.

 

Research environment and tools used in related work

The thesis can be grounded in a research environment where related projects have already used practical AI-development tools rather than purely abstract methods. The previously shared multi-agent thesis explicitly states that GitHub Copilot custom agents in VSCode were used as the implementation environment, and it also lists Python, FAISS, LangChain Core tools, and Ollama in the technology stack.

The earlier multimodal retrieval presentation additionally refers to Dataiku, LangChain, and the challenges of handling multimodal data in an industrial setting. The retrieval thesis itself discusses vector- and graph-based RAG approaches, embeddings, and hybrid retrieval strategies, which can be valuable if the log-explanation agent also needs to retrieve supporting context from surrounding documentation. In other words, the student would not be exploring an isolated problem, but one connected to a broader and already active experimentation landscape.

 

Expected outcomes

The expected outputs of the thesis include:

• a well-structured Master’s thesis report with both scientific and industrial motivation,
• a prototype or demonstrator for AI-supported log interpretation and visual engineering explanation, 
• an evaluation of the prototype in terms of technical usefulness and explanation quality,
• and recommendations for how log interpretation can contribute to the broader MONA LISA vision of AI-supported visual analytics across the engineering lifecycle.

 

All about you

We are looking for a Master’s student with interest in AI, intelligent systems, multimodal data, explanation generation, log analysis, or visual analytics. Suitable academic backgrounds include Computer Science, Artificial Intelligence, Software Engineering, Data Science, Machine Learning, or adjacent technical disciplines.

Useful qualifications include:

• solid programming ability, especially in Python, because earlier related work used Python-based pipelines and AI experimentation workflows,
• genuine interest in LLMs, agent concepts, retrieval systems, or technical data analysis, because the problem spans all of these areas,
• curiosity about how to make AI outputs more understandable and useful for engineers, because the thesis is strongly focused on explanation quality and practical value, 
• and willingness to work with a problem that combines research, prototyping, and domain understanding.

Experience with logs, data visualisation, or industrial software is an advantage, but it is not a strict requirement. A good candidate can also come from a general AI or software-engineering background and grow into the domain through the thesis work.

 

Log data is often one of the richest but also one of the most underused sources of operational insight in engineering systems. If an AI agent can help engineers interpret logs faster and more accurately, it can improve not only issue understanding but also downstream diagnosis, communication, and resolution planning. At the same time, the problem opens up important scientific questions about explanation, multimodality, retrieval, and human-centred AI support. This makes it an excellent Master’s thesis for a student who wants to build something useful while also engaging with current research in AI for engineering.

 

Application

Please apply with your CV, academic transcript, and a short motivation statement explaining your interest in the topic and any relevant background. This is a Master’s thesis opportunity intended for students who want to combine AI research with practical engineering relevance in the context of the MONA LISA initiative.

 

 

You don’t need to be a train enthusiast to thrive with us. We guarantee that when you step onto one of our trains with your friends or family, you’ll be proud. If you’re up for the challenge, we’d love to hear from you!

 

Important to note

As a global business, we’re an equal-opportunity employer that celebrates diversity across the 63  countries we operate in. We’re committed to creating an inclusive workplace for everyone.

Industry: Computer Science, Research, Technology