Physics & ML Engineer

Role overview Build predictive capabilities for industrial equipment by combining physical models with machine learning. The work spans equipment behavior, time-series sensor data, model optimization, and deployment on constrained edge devices, with input from customers about real-world machinery and failure modes.

Responsibilities - Develop physics-based models for equipment such as pumps, blowers, generators, compressors, and battery banks. - Build hybrid pipelines that combine first-principles equations with data-driven anomaly detection. - Create feature engineering workflows for time-series sensor data. - Optimize ONNX models for ARM-based edge deployment, targeting inference times of 10–50 milliseconds. - Work with industrial customers to understand equipment behavior and failure patterns. - Contribute to an open-source machine learning library.

Requirements - Strong grounding in classical physics or engineering, including areas such as thermodynamics, fluid mechanics, vibration analysis, or electrical systems. - Experience applying machine learning to time-series problems such as anomaly detection, remaining useful life prediction, or classification. - Proficiency in Python and relevant tools, including scikit-learn, PyTorch, ONNX, pandas, and NumPy. - Ability to work within edge constraints such as limited memory, CPU-only inference, and intermittent connectivity.

Nice to have - Familiarity with industrial communication protocols such as OPC UA or Modbus. - Mechanical or electrical engineering background. - A PhD is welcome but not required; specific industrial IoT experience is also not required.

Benefits and work setup - Full-time remote role with no specified time zone requirement.