Mehdi Amian

At Apple, I developed end-to-end ML pipelines for multimodal sensor data supporting health and fitness research initiatives.

I designed adaptive filtering and feature-extraction methods for physiological inference models, collaborated on study protocols and user-study analysis, and conducted failure analysis to inform next-generation sensing algorithms.

My work spans wearable PPG, ECG, IMU, and fNIRS signals, along with cardiac and brain MRI. I’ve built robust heart-rate extraction pipelines, uncertainty-aware cardiac MRI segmentation, multimodal time-series models, and Kalman-based artifact reduction for wearable data.

I also build practical ML systems, including local PDF and web RAG assistants, LoRA-fine-tuned clinical sensor summarization, and ONNX export and quantization workflows for edge deployment.