Software Engineer – Machine Learning for Edge Power Systems

About the role

We are looking for a Software Engineer specialized in Machine Learning to develop and deploy intelligent models on resource‑constrained edge hardware for powertrain applications. You will work on the full data pipeline, from sensor data acquisition to real‑time inference on embedded platforms.

Key responsibilities

• Design, train, and deploy ML models optimized for edge devices.
• Perform exploratory data analysis, statistical modeling, and visualization of time‑series sensor data.
• Engineer features and build data pipelines to prepare raw sensor datasets for modeling.
• Conduct hyperparameter tuning to improve model accuracy and efficiency.
• Optimize model inference and runtime to meet strict real‑time constraints.
• Develop real‑time application logic in C/C++ and use Python for training and preprocessing.
• Apply quantization, pruning, and memory‑optimization techniques.
• Filter and synchronize noisy sensor data using methods such as Kalman filters.
• Integrate and customize hardware platforms including Raspberry Pi, Linux/Raspbian, GPIO, and SPI.
• Test, benchmark, and validate ML estimators on real hardware or high‑fidelity simulators.

Required profile

• 3+ years of experience in machine learning, edge computing, or embedded software development.
• Proven expertise in feature engineering, dataset preparation, and data‑pipeline development.
• Strong background in applied data science, statistical analysis, anomaly detection, and time‑series visualization.
• Deep knowledge of hyperparameter tuning and low‑latency model inference.
• Hands‑on experience validating models on hardware or advanced simulators.
• Familiarity with TensorFlow Lite for Microcontrollers and Scikit‑Learn is a plus.

Required skills

• C
• C++
• Python
• TensorFlow Lite for Microcontrollers
• Scikit‑Learn
• Raspberry Pi
• Linux/Raspbian OS
• GPIO
• SPI
• Kalman filtering
• Model quantization
• Model pruning
• Hyperparameter tuning
• Time‑series analysis
• Signal processing