Skip to content
Mohamed Hafez

Mohamed Hafez

AI Researcher & Engineer

I work on efficient deep networks, deepfakes, and trustworthy AI for medical imaging, and I build the software around the models.

About

I did my MASc in Electrical and Computer Engineering at Ontario Tech University.

My research is on efficient deep networks, deepfakes and trustworthy AI, mostly on medical images. Right now I'm working on pruning ResNets, deepfake benchmarks and model calibration.

My bachelor's was in biomedical engineering, mostly working with medical data and the engineering methods behind it.

I also make short videos explaining how AI works, like forward and backward passes and encryption, on my YouTube channel The Mo Lab.

Research

  1. AI-based tool for early detection of Alzheimer's disease

    Shafiq Ul Rehman, Noha Tarek, Caroline Magdy, Mohammed Kamel, Mohammed Abdelhalim, Alaa Melek, Lamees N. Mahmoud, Ibrahim Sadek

    Heliyon 10(8), 2024 · listed as Mohammed Abdelhalim

    A VGG16 that only looks at hippocampus slices reaches 98.17% test accuracy on CN / MCI / AD (ADNI). We also built BrainAware, a web app around it.

In progress

  • LAGD-Net: Differentiable Block Gating for Depth Pruning of Residual Networks. Learned, temperature-annealed gates drop whole residual blocks during training. On CIFAR-10 ResNet-101 it removes 14 of 33 blocks for a 1.63× speedup at −0.17 points.
  • Block-selection benchmark for depth pruning. Which blocks to remove, and does it matter after fine-tuning? 8 architectures, 3 datasets, 740 runs.

All papers on Google Scholar

Projects

  • BioMed Lab

    15 of my old biomedical desktop apps rebuilt as browser tools: CT reconstruction, an MRI viewer with 3D, ECG aliasing, Z-transform filter design, sequence alignment and more. Everything runs locally.

    React · TypeScript · Web Workers

  • Auto-Med-Agent

    Medical research assistant in one chat. Questions go through RAG over medical papers (MPNet + FAISS retrieval, cross-encoder re-ranking, RoBERTa extractive QA, BART summaries); PubMed and ClinicalTrials.gov queries go to their APIs.

    Transformers · FAISS · Streamlit

  • AI Resume Screener In progress

    Parses resumes locally or with a Groq-hosted LLM, scores them against a job with spaCy + Sentence-BERT, and ranks applicants. Backend works end to end; the dashboards are next.

    Built with Hesham Atef.

    FastAPI · PostgreSQL · Next.js

  • BrainAware

    Web app around the Alzheimer's model: upload a NIfTI or DICOM brain MRI, get CN / MCI / AD probabilities, view the scan in 2D and 3D, and track each patient over time.

    Django · React · vtk.js · TensorFlow

Smaller things

More on GitHub

Background

Experience

  • Vector Institute

    Starting Nov 23, 2026

    Associate Machine Learning Specialist

  • Vector Institute

    May 2026 – Nov 2026

    Applied Machine Learning Intern

    MOZAIC audio-video deepfake benchmark.

  • Ontario Tech University

    May 2025 – Nov 2026

    Teaching Assistant

  • Cairo University

    Sep 2023 – Jan 2025

    Teaching Assistant

  • Webbing Stone

    Jun 2021 – Oct 2022

    Flutter Developer (part-time)

  • Medical Equipment Calibration Lab

    Jun 2020 – Sep 2020

    Intern

Education

  • Ontario Tech University

    May 2025 – Oct 2026

    MASc, Electrical and Computer Engineering

    IoT Research Lab, supervised by Prof. Khalid Elgazzar.

  • Cairo University

    2018 – 2023

    BSc, Biomedical Engineering

    Faculty of Engineering. Mostly medical data and the engineering methods behind it.