Data Science | AI Engineering internship: Stochastic optimization for value-driven Predictive Maintenance

ASML maintains an extensive collection of diagnostic documents called PCS that detail the Problem, Cause, and Solution faced by our engineers while Installing or Servicing the machines. While these documents contain valuable diagnostic information, their unstructured format makes it challenging to scale and standardize diagnostic procedures. A better alternative is an executable diagnostic process called DDFs which offers a more consolidated and structured approach by using dynamic flowcharts to guide engineers through a sequence of actions to identify the failure cases.  As our install base and engineer workforce increases, providing service actions in a guided way compared to relying on the engineer's ability to find the correct PCS becomes even more significant. Leveraging our existing PCS to DDF can also help us unlock valuable data mining opportunities, enabling us to understand which PCSs are more used, successful, or relevant ultimately leading to improved and more leaner knowledge base management.

Your assignment

This internship project focuses on developing an intelligent system to automate the conversion of legacy Problem-Cause-Solution (PCS) documents into structured Deterministic Diagnostic Flow (DDF) documents using LLMs. The project aims to streamline diagnostic processes while leveraging the existing organizational knowledge.

Generate diagnostic questions and service action from PCS documents using LLMs. Compare LLMs performance against other baseline models. If necessary, fine tune LLMs using ASML specific data to improve performance. Evaluate the LLMs performance for the question generation problem using standard evaluation techniques. Scaling system to handle large number of PCS documents. Stretch Goal: Constructing control matrix on the questions generated to lead to unique failure modes.

This is a master thesis internship for a minimum of 6 months, for 4 to 5 days per week (at least 3 days onsite). The start date of this internship is as of September 2026, but earlier is possible.

Your profile

To be a great match for this internship you:

• Are a graduating master student in Computer Science, Data Science, Machine Learning or a similar field.

• Are enthusiastic about GenAI and LLMs or related technologies.

• Have some basic knowledge of or familiarity with PyTorch and/or TensorFlow and are experienced in Python.

• Are able to work independently and autonomously and are pro-active.

• Have strong communication skills and are fluent in English (verbal and written).