Molex is building a workflow that turns engineering simulation results into fast, physics-aware predictions. In this onsite role in Lisle, IL 60532, you will design and operationalize physics-informed surrogate models that can screen candidate designs in milliseconds, helping accelerate design optimization.
You will work across model development and production deployment on Azure Machine Learning, partnering with data scientists and MLOps teams to ensure models are versioned, monitored, benchmarked, and continuously improved as new simulation outcomes become available.
What you will do
- Design and train surrogate models, including neural networks, Gaussian processes, gradient-boosted trees, and GNNs/PINNs, using Azure GPU compute (ND/NC series).
- Incorporate physics-informed constraints so predictions remain physically valid rather than only matching statistical patterns.
- Develop model-uncertainty and confidence scoring to determine which designs require full simulation validation, then retrain as new results arrive.
- Deploy, version, and manage models through Azure ML endpoints and the model registry, with rolling drift monitoring.
- Benchmark surrogate performance against full-simulation to quantify speedups and inform platform-level performance tuning.
Requirements
- Extensive hands-on experience building, training, and deploying ML models in production, not limited to using pretrained APIs.
- 10+ years building ML for physical or engineering systems, including surrogate modeling, physics-informed ML, or scientific ML.
- Strong Python skills with PyTorch or TensorFlow.
- Understanding of relevant engineering or physics fundamentals and the simulation data formats for your domain.
- Experience with Azure Machine Learning or a similar cloud ML platform.
- Familiarity with uncertainty quantification, including Bayesian approaches and ensembling.
Technologies
- Azure Machine Learning; Azure GPU compute (ND/NC series)
- Python; PyTorch; TensorFlow
- Azure ML endpoints; model registry
- Neural networks; Gaussian processes; gradient-boosted trees; GNNs; PINNs
- Bayesian approaches; ensembling; GPU-heavy training
Benefits
- Medical, dental, and vision
- Flexible spending and health savings accounts
- Life insurance, disability, and retirement
- Paid vacation/time off
- Educational assistance
- May also include infertility assistance
- Paid parental leave and adoption assistance
Additional experience that can put you ahead: Direct experience with industry-standard EM or physics simulation tools; geometric deep learning for CAD data (graph neural networks, mesh-based models); background in RF/high-speed electronics or interconnect design.
Location: Lisle, IL 60532 (onsite). Compensation: USD 170,000 - 250,000 per year. Experience level: 10+ years.
