Machine Learning Engineer — Industrial Quality Systems

Mei Lin

Contract · Mid-level · Zurich

This role is no longer accepting applications

Similar open roles you can apply to now:

About the role

We are a Swiss industrial technology company. We have manufactured precision measurement and inspection equipment since 1954. Our instruments are used in semiconductor fabrication, medical device production, aerospace component validation, and pharmaceutical quality control. Our customers include TSMC, Medtronic, Airbus, and Pfizer. Precision is not a value we aspire to. It is the minimum standard from which all work begins. We are hiring a Machine Learning Engineer on a 12-month contract to work within our R&D division on the application of ML-based image analysis to our inspection systems. Our instruments produce high-resolution images of manufactured components. Currently, defect detection is performed through rule-based thresholding algorithms developed over many years. We have a research hypothesis — that ML-based vision models can detect subtle defect patterns that the rule-based system misses, at production throughput speeds, with a false positive rate that does not introduce unacceptable rework costs. Your role is to test that hypothesis rigorously, implement a prototype system, and provide us with a clear technical assessment of whether it is ready for integration into a production instrument.

Responsibilities

Design and train ML models for automated visual defect detection on precision component inspection images
Conduct systematic evaluation against existing rule-based system benchmarks — statistical rigour in the comparison is non-negotiable
Optimise inference performance for real-time operation at production throughput rates
Produce detailed technical documentation of every experiment, model decision, and evaluation methodology
Provide a written technical assessment at the end of the engagement with a clear recommendation on production readiness

Requirements

4+ years of ML engineering with a focus on computer vision or image-based defect detection in industrial or scientific contexts
PyTorch for training and evaluating image classification and anomaly detection models on domain-specific industrial datasets
Scikit-learn for baseline model development, feature engineering, and evaluation methodology
Python at a professional engineering standard — reproducible experiments, documented code, and version control discipline
MLflow for experiment tracking, model versioning, and maintaining the reproducibility of every result you produce
Docker for reproducible development environments that can be handed off cleanly at the end of the contract
Experience with high-precision evaluation: understanding of precision-recall curves, statistical testing of model improvements, and the difference between laboratory accuracy and production reliability

Benefits

CHF 130,000 – CHF 158,000 pro-rated annual contract rate
12-month initial contract — extension possible if integration phase proceeds
Remote-first with two visits per quarter to our Zurich R&D facility — travel and accommodation covered
Access to proprietary industrial inspection datasets unavailable in academic or open source contexts
Work on a system where your evaluation methodology will itself be subject to Swiss engineering rigour — which we appreciate you will find either appealing or disqualifying