Data Science Internship: Improving Machine State Classification with Machine Learning

Introduction

ASML derives machine performance and availability information from machine-generated data using a standardized State Model, aligned with the SEMI E-10 standard.

The reconciled states are used for reporting, performance metrics, and contractual purposes. While the automated State Model provides a consistent first interpretation, reconciliation remains necessary to incorporate business and operational context that is not fully available in machine data alone.

Despite continuous improvements to the State Model, differences between automatically generated states and reconciled states still occur. These differences typically originate from ambiguous task sequences, Limited interpretation of test activities and Operational intent that is documented in logbooks but not reflected in event logs. As a result, reconciliation requires significant manual effort and introduces delays before final machine states are available.

The focus of this internship is to improve the State Model using Machine Learning.

Your Assignment

The objective of this assignment is to investigate whether Machine Learning techniques can be applied to:

• Improve the quality of automatically generated machine state.
• Reduce the number of corrections required during reconciliation.
• Increase consistency across machines, sites, and product lines.

Your responsibilities include:
1. Understanding the Existing Pipeline

• Event logs
• Task interpretation
• State Model logic (states, substates, triggers)
• Reconciliation process and EPC workflow

2. Dataset Preparation

• Alignment of original states and reconciled states
• Identification of reconciliation changes
• Feature extraction from event logs, machine tasks and task transition, and selected information from logbooks (subject to availability)

3. Machine Learning Model Development

• Supervised learning using reconciled states as ground truth
• Prediction of machine states and/or state transitions
• Justification of model choice and features

4. Evaluation

• Comparison of Current State Model output, ML-based predictions and reconciled states
• Quantitative assessment of potential quality improvements

5. Conclusions and Recommendations

• Feasibility of ML support for state interpretation
• Possible integration approaches (decision support vs automation)
• Limitations, risks, and explainability considerations

This is a Master (thesis) internship for minimum 6 months, minimum 4 days per week (minimum 2 days on-site). The start date of this internship is as of September 2026.

Your Profile

To be a good match for this internship, you:

• Are a master student in Data Science or a related field.
• Have experience with programming (preferably Python).
• Have strong analytical skills and are a structured thinker.
• Are proactive and have the ability to take ownership.

Other requirements you need to meet:

• You are enrolled at an educational institute for the entire duration of the internship;
• Attach your cover letter with a clear motivation why you are interested in this internship assignment in particular;
• You need to be located in the Netherlands to perform your internship. In case you are currently living/studying outside of the Netherlands, your CV/motivation letter needs to include the willingness to relocate.
• If you are a non-EU citizen, studying in the Netherlands, your university is willing to sign the documents relevant for doing an internship (i.e., Nuffic agreement).

Inclusion and diversity

ASML is an Equal Opportunity Employer that values and respects the importance of a diverse and inclusive workforce. It is the policy of the company to recruit, hire, train and promote persons in all job titles without regard to race, color, religion, sex, age, national origin, veteran status, disability, sexual orientation, or gender identity. We recognize that inclusion and diversity is a driving force in the success of our company.

Need to know more about applying for a job at ASML? Read our frequently asked questions.