In short
Machine Learning Engineer at North Sails (global sailmaker) building, training, and benchmarking ML surrogate models that emulate aerodynamic physics for sail design. Day-to-day involves Python/PyTorch or TensorFlow model development, CFD data pipelines, experiment tracking, and validation with designers. Remote-first across UK, France, Italy, Spain, Netherlands.
About the role
ML Engineer – Surrogate Modelling
📍 Location: Hybrid (UK, France, Italy, Spain, Netherlands)
💼 Employment Type: Full-time, Permanent
🎯 Level: Early Career / Mid-Level
Company Overview
North Sails is the global leader in sailmaking, designing and manufacturing sails for premier racing and cruising applications worldwide, including the America’s Cup, the Ocean Race, and Grand Prix racing circuits. Design and production operations are integrated across international hubs.
Program Overview
Sail design presents a complex coupled fluid-structural challenge. A flexible sail deforms under aerodynamic loading, causing its operational flying shape to differ significantly from its manufactured geometry. To evaluate high-dimensional design spaces at speed, machine learning surrogate models are being integrated directly into the design pipeline. The primary objective is developing surrogate models capable of rapidly predicting integrated aerodynamic forces, moments, and surface pressure distributions for user-defined flying shapes and operational conditions.
Position Summary
The Machine Learning Engineer – Surrogate Modelling focuses on constructing, training, and systematically evaluating machine learning models configured to emulate complex aerodynamic physics. Working closely with lead engineers and domain designers, the engineer conducts empirical model diagnostics, benchmark testing, and feature engineering to ensure surrogate accuracy meets stringent engineering design requirements.
Key Responsibilities
- Develop, train, and benchmark machine learning surrogate models utilizing computational aerodynamics datasets.
- Execute structured experiment suites across model architectures, hyperparameter spaces, and feature representations with rigorous tracking.
- Construct evaluation frameworks, out-of-sample validation tests, and designer reference benchmarking suites.
- Investigate model failure modes to isolate underlying causes across data quality artifacts, architectural limits, or physical boundary edge cases.
- Preprocess and format raw high-dimensional simulation data into robust training data pipelines.
- Maintain complete documentation of experimental setups, hyperparameter sweeps, and validation results.
- Engage directly with sail designers to analyze surrogate predictions against domain intuition and high-fidelity baseline data.
Qualifications & Requirements
- Core foundation in machine learning principles supported by 2+ years of practical application or an applied postgraduate degree (MSc/PhD).
- Proficiency in Python programming and modern deep learning frameworks (PyTorch or TensorFlow) using structured software design principles.
- Academic or industrial background in physics, mathematics, computational engineering, or related quantitative disciplines to reason effectively about simulation data.
- Analytical problem-solving capability focused on root-cause failure analysis in trained models.
- Clear technical documentation and communication habits.
Good to Have (Desirable)
- Exposure to scientific machine learning, physics-informed neural networks (PINNs), operator learning, or geometric ML.
- Familiarity with simulation technologies (CFD/FEA) or experiment tracking software (e.g., MLflow, Weights & Biases).
- Interest in fluid dynamics, aerodynamics, or marine engineering.
Working Arrangements & Employment Conditions
This position operates on a remote-first model with eligible employment locations in the United Kingdom, France, Italy, Spain, or the Netherlands. Periodic travel is required for engineering and design integration reviews. Compensation is aligned with regional market benchmarks in the country of employment.