Hi, I’m Shervin. I’m an applied mathematician and engineer with a focus on machine learning. I grew up between a small town by the sea and Tehran, and now live in Canada.
I studied electrical engineering at Sharif University of Technology, where I worked on EEG based brain computer interface systems for wheelchair control. Later, at Mindpax, I worked on time series models for mental health data. At the University of Waterloo, I worked on kidney modeling and machine learning using large scale diabetes data. I now work at Numeris, building models with behavioral and audience data.
This site is a little more personal than a resume. It’s where I share projects, notes, and some of my writing.
Outside work, I read, write, take photographs, learn languages, stay active, and travel. I’ve spent a good amount of time traveling on my own, and adventuring. I speak English, Persian, German, and French, and hoping to learn more.
Selected Projects and Publications
Mechanistic Modeling of Renal Drug Response
Built a steady state nephron model that includes calcium and magnesium transport to study how drug induced transporter changes alter electrolyte balance along the kidney.
Robust and Interpretable Deep Learning for Genomic Data
Built adversarially trained deep learning models for genomics to test sensitivity of gene expression predictions to small perturbations and improve biological interpretability for drug discovery.
Unsupervised Modeling of Long Horizon Behavioral Time Series
Built unsupervised time series pipelines to identify recurring long term trajectory patterns in irregular multi channel behavioral data using DTW, representation learning, and density based clustering.
EEG Driven Brain Computer Interfaces for Assistive Mobility
Designed and implemented an end to end EEG based BCI for real time wheelchair control using CSP feature extraction and frequency domain classification to translate neural signals into reliable movement commands.