Graduation​/internship Controls Engineering | Design LLM-driven 'Virtual Site Engineer

Position: Graduation/internship Controls Engineering | Design LLM-driven 'Virtual Site Engineer'
Downtime of production lines is costly. Traditional maintenance strategies often only react after a failure has occurred. Using LLMs, we can interpret error messages and sensor data in real-time to provide proactive maintenance advice. The intern/graduate student designs and builds a prototype of a 'Virtual Site Engineer': an AI agent that continuously analyses operational and semantic data to detect anomalies, identify causes, and suggest action plans.

This project is intended for two candidates and covers both the IT and mechatronics aspects. For the software side (databases, LLMs, and interface software), we have already found a suitable student. We are still looking for someone with a stronger focus on machine control, mechatronics, and industrial automation. There is some overlap between these two disciplines.

Assignment description
The intern/graduate student designs and builds a prototype of a 'Virtual Site Engineer': an AI agent that continuously analyses operational data (sensor data, alarms) and semantic data (error logs, manuals). The system must be capable of detecting an anomaly, identifying the cause, and proposing an action plan.

Concrete goals

Data integration:
Setting up a pipeline to access real-time and historical data, for example via OPC UA, MQTT, or directly from an MES/ERP system;

Contextualisation:
Implementing a RAG architecture that supplies the LLM with the appropriate context (machine manuals, previous error reports) to increase diagnostic accuracy;

Action generation:
Developing a mechanism (e.g., an agentic workflow) that generates a sequence of actions based on the diagnosis, ranging from a simple notification to a safe machine command

Expected results

A working prototype capable of monitoring one of RIWO's machines;

A detailed design report on the chosen architecture and technologies;

An evaluation of diagnostic quality and a cost-benefit analysis for the client.

Technical requirement & challenges

Integration with industrial communication protocols (OPC UA, MQTT);

Processing of both structured (sensor signals) and unstructured data (logs, PDFs);

Design of safety mechanisms (guardrails) to prevent the AI from performing harmful actions.

What will you learn?

Knowledge of AI-driven predictive maintenance (PdM) and prescriptive maintenance;

Experience with building 'agentic' workflows;

Personal (SMART) learning objectives that contribute to obtaining your diploma;

Insight into ensuring safety and reliability in AI applications.

Stake holders

Team leader Controls Engineering

Systems Engineer

Technical Manager

The Team leader Controls Engineering, Marc Schothorst, is your point of contact.

About The Project
Both graduation and internship are possible. The scope and content of the project can still be tailored to meet the requirements of your academic program.

Graduating / internship at RIWO
We are delighted by curious, enthusiastic and independent students who want to look back on a meaningful internship. With us you get a lot of freedom to develop yourself, with personal guidance. Take control of your assignment, tailor it yourself. Your ideas are welcome!

Do you have any questions or would you like more information? Please contact Marc Schothorst, team leader Controls Engineering via m.schothorst.

#J-18808-Ljbffr