Research & Publications

My research life started almost by accident: my master's project, at the University of Mazandaran, was on online signature verification for the Persian language. Building that system — the feature engineering, the dataset, the sheer amount of programming — is where machine learning really grabbed me.

That interest carried me to Polytechnique Montréal, where I did my PhD in Computer Engineering as a member of the Liv4D lab, supervised by Prof. Farida Cheriet. My focus was interpretable deep learning for diagnosing acute respiratory distress syndrome (ARDS) in children from chest X-rays — part of a project with Sainte-Justine, the pediatric hospital in Montréal, working closely with my clinical supervisor, Dr. Philippe Jouvet. I wanted models that were not just accurate, but explainable enough for clinicians to actually trust, on a problem where the patients were kids.

These days I'm a Machine Learning Engineer at Junction AI, where I build LLM and RAG systems — combining semantic and keyword retrieval, chain-of-thought prompting, and solid evaluation tooling, in Python and PyTorch, with vector databases. What connects all of it, from the hospital to industry, is the same goal: AI that's genuinely helpful and reliable, not just clever in a demo.

Selected publications

Here's my publication record, grouped by theme. According to Google Scholar it adds up to around 155 citations and an h-index of 5 — modest, but each one represents a problem I genuinely cared about.

Medical imaging / ARDS (from my PhD):

  • Dense-Unet: a light model for lung fields segmentation in Chest X-Ray images — M Yahyatabar, P Jouvet, F Cheriet — IEEE EMBC, 2020. This is my most-cited paper.
  • A web-based platform for the automatic stratification of ARDS severity — M Yahyatabar, P Jouvet, D Fily, J Rambaud, M Levy, RG Khemani, et al. — Diagnostics 13(5), 2023. This is the tool I'm proudest of — it was built to actually help children.
  • Joint classification and segmentation for an interpretable diagnosis of ARDS from chest x-rays — M Yahyatabar, P Jouvet, F Cheriet — Journal of Medical Imaging 10(5), 2023.
  • PhD thesis: Interpretable Learning Models for Acute Respiratory Distress Syndrome Detection From Chest Radiographs — École Polytechnique de Montréal, 2023.

Online signature verification (from my master's, Persian-language focused):

  • Online signature verification using double-stage feature extraction modelled by dynamic feature stability experiment — ME Yahyatabar, J Ghasemi — IET Biometrics 6(6), 2017.
  • Online signature verification: A Persian-language specific approach — ME Yahyatabar, Y Baleghi, MR Karami — ICEE, 2013.
  • Online signature verification: a robust approach for Persian signatures — ME Yahyatabar, Y Baleghi, MR Karami — JIST 3(2), 2015.

Recent LLM & security (co-authored):

  • Advanced smart contract vulnerability detection using large language models — F Erfan, M Yahyatabar, M Bellaiche, T Halabi — CSNet, 2024.

  • Fine-Tuned Large Language Model for Securing Ethereum Smart Contracts with Real-Time VSCode Auditing — F Erfan, M Yahyatabar, M Bellaiche, T Halabi — Blockchain: Research and Applications, 2026.

  • My full profile on Google Scholar