Machine Learning Engineer

ABOUT THE PROJECT

An R&D project extending an industrial equipment monitoring system with AI, to detect unusual

behaviour that fixed rules and thresholds cannot capture, including on complex equipment such as pumps and motors in rolling mills.

A working prototype already stores data, trains models and runs predictions. The goal is to develop it and validate it on real industrial cases, focusing on:

▪ Anomaly detection on large volumes of sensor data.

▪ Production context , so normal variations (e.g. higher current draw when processing harder

material) don't trigger false alarms.

▪ Explainable alerts that point to the relevant signals and, potentially, use an LLM with technical documentation to suggest likely causes.

▪ Wider analysis over time, from time series to logs, images and video.

THE ROLE

We are looking for a Machine Learning Engineer with strong software development skills to work on an existing prototype and test solutions on real industrial data. The role combines application development with data analysis and model experimentation.

As this is an R&D project, a significant part of the work is experimenting, comparing methods and deciding, based on results, which approach fits each type of equipment.

The collaboration is a B2B type of contract and it's estimated until April 2027.

RESPONSIBILITIES

▪Understand the existing prototype's codebase, extend it with new features, and keep it clean,

testable and maintainable.

▪ Build processing pipelines for sensor data: synchronising multiple signals, handling missing

values and noise, time windows, high-frequency sampling.

▪ Select, implement and compare statistical and machine learning methods for anomaly

detection, including on unlabelled data or data that already contains anomalies.

▪ Model the production context (materials, operating regimes) to separate normal variations from real equipment issues.

▪ Define the validation methodology: train/test splits without leakage, relevant metrics, tracking false alarms and missed anomalies.

▪ Develop explainability mechanisms that identify the signals behind an alert and present results in a form useful to operators.

▪ Integrate training and inference into the application, ensuring experiment reproducibility and

model versioning.

▪ Explore, together with the team, the use of LLMs and technical documentation to suggest

possible causes of an alert.

▪ Document experiments and results as a basis for the project's technical decisions.

REQUIREMENTS

▪ Hands-on experience as a Machine Learning Engineer, production-oriented Data Scientist, or

Software Engineer with delivered ML projects.

▪ Strong software development skills, preferably in Python: structured, tested code, version

control (Git), working in an existing codebase.

▪ Experience with multivariate time series: synchronisation, missing values, noise, time windows, high-frequency data.

▪ Knowledge of statistical and ML anomaly detection methods, and the ability to choose and

compare them based on the data, including in unsupervised settings.

▪ Experience processing large data volumes efficiently without loading the full dataset into

memory (chunked processing, storage formats suited to time series).

▪ Solid understanding of model evaluation: validation strategies for temporal data, avoiding data leakage, balancing false alarms against missed anomalies.

▪ Experience with model explainability techniques and presenting results to non-technical users.

▪ Experience integrating models into applications: training and inference pipelines, experiment

reproducibility, model versioning.

▪ Analytical mindset and appetite for experimentation, suited to an R&D environment.

NICE TO HAVE

▪ Integrating LLMs via API and using RAG to query technical documentation.

▪ Experience with logs, images or multimodal models.

▪ Experience in condition monitoring, predictive maintenance or industrial processes.

▪ Using AI coding tools, while reviewing and understanding the generated code.