AI & Software Engineer · Open to work

Hi, I'm Ahmed Elbadawy Mohamed.

I am an AI engineer (MSc, Western University) specializing in deep learning and medical image analysis, with hands-on experience across the full stack, from model development to embedded and web applications.

Personal Photo

A bit about me

I'm an AI engineer and Software developer with a master's in Machine Learning for Health and Biomedical Sciences from Western University and a background in Systems and Biomedical Engineering. My work spans deep learning and medical image analysis, but what I enjoy most is the full arc, taking a model from research into web systems that actually put it to use.

I care about building things that work in real-world settings, under real constraints like noisy data and limited datasets. Coding and AI are what let me turn ideas into real products with real impact, and I love that — along with the constant learning the field demands.

Outside of work, I'm usually gaming, watching movies, or out running. I also love to travel and camp, discovering new places is one of my favorite ways to reset.

Education

MSc, Machine Learning in Health & Biomedical Sciences @ Western University Expected 2026

London, ON, Canada

BSc, Systems & Biomedical Engineering @ Cairo University 2022
  • Graduated with Excellent honours (3.83 GPA), ranked 3rd of 105 in the final year.
  • Thesis (ranked 2nd in department) on the role of DNA-repair-gene mutations in cancer evolution.

Where I've worked

Research Assistant @ Robarts Research Institute, Western University 2024 — Present · London, ON
  • Built a residual U-Net to refine noisy CT perfusion maps from lower-temporal resolution CT scans, using initial physics-based perfusion outputs as input and high-temporal-resolution maps as ground truth; applied pretraining and a lightweight architecture to handle a limited dataset.
  • Developing a segmentation model to delineate ischemic penumbra and core infarct on cerebral perfusion maps, computing the mismatch ratio used to determine stroke patients' eligibility for endovascular thrombectomy (EVT).
  • Built a single-acquisition pipeline using deformable image registration, signal processing, and physiological signal decomposition to generate lung ventilation and perfusion maps from one CT scan, validated against an independent reference modality.
AI Research Assistant @ Sejong University Mar 2024 — Sep 2024 · Seoul, South Korea

Built a multi-stage deep-learning pipeline to detect and grade lumbar spinal stenosis from MRI.

  • Used a YOLO model to localize the region of interest on lumbar MRI scans.
  • Applied an EfficientNetV2 CNN to grade stenosis severity (normal, mild, moderate, severe).
  • Integrating an LLaMA-based LLM to generate text explanations justifying each grading decision, adding explainability to the pipeline.
Clinical Engineer (Mandatory Military Service) @ Elhelmia Military Hospital Jan 2023 - Feb 2024 · Cairo, Egypt
  • Performed preventive maintenance and functionality checks on medical equipment, including defibrillators, ECG machines, and ventilators, and resolved minor issues with CT workstations and other devices.
  • Coordinated with vendor service teams for specialized maintenance and handled the department's routine clinical-engineering administration.
AI Research Intern @ Valeo Sep 2022 - Jan 2023 · Giza, Egypt

Explored using LLMs to automate unit-test generation for embedded developers, aiming to save developer time.

  • Fine-tuned a CodeT5 (T5-architecture) model on C code to automatically generate unit tests for C methods; produced a working proof-of-concept.
  • Conducted a literature review on LLM-based test generation and presented the approach and its potential value to the team.
Bioinformatics Research Intern @ Centre for Genomic Regulation (CRG) Aug 2021 - Jul 2022 · Barcelona, Spain (Remote)

Investigated the role of DNA-repair-gene mutations in cancer evolution, combining deep learning, population-genetics simulation, and large-scale data engineering.

  • Applied deep-learning–based mutational signature decomposition to trace mutation sources and quantify the proportion driven by deficient repair genes.
  • Built Wright–Fisher simulations to test whether tumorigenesis depends on repair-gene mutations (mutation-rate modifiers) or occurs without them, modeling how these mutations raise the overall mutation rate and the likelihood of acquiring driver mutations.
  • Engineered Python data-processing pipelines on Azure to process 3M+ mutational records.

Projects

Lumbar spine MRI with colored segmentation overlay

Lumbar Spine MRI Segmentation

A U-Net with Squeeze-and-Excitation attention that segments five lumbar spine structures from axial MRI to support diagnosis of lumbar spinal stenosis. Tuned via Weights & Biases hyperparameter sweeps, it reaches a 92.78% test Dice score — outperforming the published benchmark on the dataset.

Python PyTorch MONAI Albumentations W&B
SIFT features example

Computer Vision Library

A computer vision library implementing classical CV algorithms entirely from scratch, without high-level frameworks. It spans edge and feature detection (Canny, Hough, Harris, SIFT), segmentation (K-Means, Mean-Shift, Otsu), and a complete face detection and recognition pipeline. Includes an interactive interface for visualizing and testing each algorithm in real time.

Python C++
Car app fig

Self-Driving Car

A self-driving car built on Arduino and ESP32, controllable through a Progressive Web App over WiFi. An onboard camera runs OpenCV lane detection that steers the car autonomously via rule-based control, and users can switch between manual and self-driving modes from the app.

Python OpenCV Arduino ESP32 Node.js PWA
Web app fig

Z-Plane Digital Filter Visualizer

An interactive browser-based tool for designing discrete-time digital filters by placing poles and zeros on the Z-plane. Submit your design to instantly visualise the magnitude and phase frequency response, with optional all-pass phase correction.

Python Flask chart.js
Web app fig

Hemodialysis Clinic Web App

A full-stack web application for managing a hemodialysis clinic, with role-based access for patients, physicians, and admins. Patients book appointments (synced to Google Calendar via API) and receive their scans and medical reports online; physicians manage appointments, share reports with patients, and collaborate through a built-in blog; admins have full control over users, content, and clinic statistics.

Python Flask SQLAlchemy HTML CSS
Sound equlizer app example

Sound Equalizer

A multi-document desktop DSP application for real-time audio equalization across 10 frequency bands. Users can play audio, view and animate the signal before and after equalization, and explore a customizable spectrogram — with all slider settings saved automatically per file.

Python PyQt
Song recog app example

Song Recognition

An audio fingerprinting app that identifies songs by their spectral signature. It extracts mel-spectrogram and MFCC features, hashes them perceptually, and matches a query against a database using Hamming distance to rank the closest songs. Also supports blending two songs by a weighted average to find what the mix most resembles.

Python PyQt
Articulated 3D Human Figure

Articulated 3D Human Figure

An interactive 3D articulated human figure built with OpenGL and GLUT. Every joint in the body is independently controllable via keyboard, with full 3D camera navigation using mouse drag and arrow keys.

C OpenGL GLUT Camera mathematics
Articulated 3D Human Figure

Elevator Control System

A 5-floor elevator controller built on the C8051F020 microcontroller, simulated in Proteus. Handles real-time floor requests from both inside the cabin and hall call buttons, with a servo-driven door and floor display.

Embedded C STM8051 Proteus
Pipeline fig

EEG Seizure Detection

A machine learning pipeline that detects seizure activity in EEG recordings (CHB-MIT dataset). It combines signal preprocessing and frequency/statistical feature extraction with PCA and an SVM classifier to distinguish ictal from preictal states.

Python scikit-learn PyEEG

Publications

Evaluating AI-Powered Predictive Solutions for MRI in Lumbar Spinal Stenosis: A Systematic Review 2025

Artificial Intelligence Review (journal)

AI Segmentation Framework for Identifying Pathological Areas Causing Lumbar Spinal Stenosis 2024

IRES International Conference Proceedings, Jeju Island

Presentations

CT Perfusion for Simultaneous Imaging of Lung Ventilation and Perfusion 2026

Oral presentation at SPIE Medical Imaging and AAPM (Vancouver) · Science Council Innovation in Medical Physics finalist · Manuscript in preparation

Multiphase Coronary CTA-Based Myocardial Perfusion Estimation Using U-Net Refinement 2026

Accepted oral presentation at AAPM Annual Meeting (Vancouver) · Manuscript in preparation

Skills

Languages

Python C++ Java MATLAB R JavaScript

ML & Data

PyTorch TensorFlow scikit-learn PySpark Weights & Biases Power BI Snowflake

Web

React Angular Node.js Flask HTML CSS

Databases

SQL MySQL SQLAlchemy

Cloud & DevOps

AWS Azure Docker Git

Embedded & Other

C Arduino ESP32 Assembly (8051) OpenGL PyQt

Let's talk

I'm currently open to AI & software roles. The fastest way to reach me is email, I usually reply within a day.