BIOMEDICAL ENGINEERING — BLIDA, ALGERIA

I keep medical
imaging systems running,
and teach them to see.

Beloufa Dania — Biomedical Engineer with hands-on experience maintaining CT, MRI, and radiotherapy equipment in clinical settings, and a deep learning researcher building diagnostic AI for pediatric chest imaging.

See the research Download CV
99.4% peak model accuracy
pediatric chest X-ray classification
8 CNN architectures
benchmarked end-to-end
2.5yrs clinical biomedical experience
military hospital, Algiers
5 imaging modalities
CT · MRI · radiotherapy · nuclear medicine · dental
01 — PROFILE

Most engineers choose a side: the hardware in the radiotherapy suite, or the model that reads the scan. I work across both, because in practice they're the same problem — keeping diagnostic information trustworthy, from the machine that captures it to the algorithm that interprets it.

I'm a biomedical engineer with a Master's degree in Instrumentation Electronics from Saad Dahleb University, Blida. For close to two years, I worked inside the Direction des Équipements et de la Maintenance Technique at a major military hospital in Algiers — maintaining and calibrating CT scanners, MRI units, linear accelerators, nuclear medicine systems, and dental imaging equipment, and documenting faults with the same rigor I'd later apply to model evaluation metrics.

That clinical grounding is what shaped my Master's thesis: an AI system that classifies pediatric chest X-rays for pneumonia, built and benchmarked across eight CNN architectures. I care about models that are accurate and deployable in real, resource-limited hospital settings — not just impressive on a clean benchmark.

ArabicNATIVE
EnglishFLUENT
FrenchFLUENT
02 — EXPERIENCE

Where the equipment lives

Biomedical Engineering Intern

OCT 2022 — MAR 2025

Military Hospital (HCA Dr. Mohamed Seghir Nekkache), Ain Naâdja, Algiers

  • Maintained, calibrated, and supported clinical use of CT, MRI, radiotherapy (linear accelerators, TomoTherapy/Radixact), nuclear medicine, and dental imaging systems.
  • Diagnosed and resolved equipment faults — including linear accelerator control errors, ion-pump failures, and treatment-application faults — and documented corrective actions for traceability.
  • Worked alongside engineers and technicians across ophthalmology, ENT, radiology, and oncology to evaluate equipment performance against safety standards.
Preventive Maintenance Calibration Compliance Documentation

Teaching Assistant — Biophysics

OCT 2025 — JAN 2026

Saad Dahleb University, Blida 1 — Bachelor level

  • Supervised practical sessions using computational tools (SRIM) for undergraduate biophysics students.
  • Guided students through scientific reasoning, data interpretation, and experimental analysis.
Biophysics SRIM Mentoring

Operations Manager (Part-Time)

2024 — 2025

Patisserie Flora, Blida — leadership, organization, and decision-making while balancing technical study.

Team Member (Part-Time)

2022 — 2023

THE HIVE Coffee Shop, Blida — communication, adaptability, and teamwork under pressure.

M.SC · 2025

Instrumentation Electronics

Saad Dahleb University, Blida 1 — Instrumentation, Medical Imaging, Signal Processing, Artificial Intelligence

B.SC · 2023

Biomedical Engineering

Saad Dahleb University, Blida 1 — Signal Processing, Radiology, Biomedical Devices, Biophysics

03 — FEATURED RESEARCH

PneumoCNN4

A deep learning system for classifying pediatric chest X-rays — Master's thesis, defended June 2025.

Pneumonia kills more than 700,000 children under five every year. In Algeria, acute respiratory infections account for nearly a quarter of pediatric hospitalizations — and diagnosis is often delayed by image variability, a shortage of specialist radiologists, and hospital pressure during outbreaks. PneumoCNN4 is a proof of concept for automated, fast, and reliable triage support, designed to function even in modestly equipped hospitals.

01

Dataset

5,856 pediatric chest radiographs, normal vs. pneumonia, across three explored train/validation/test splits.

→
02

Preprocessing

Resized to 224×224px, normalized, and augmented (rotation, zoom, horizontal flip) to improve generalization.

→
03

Transfer Learning

8 ImageNet-pretrained backbones fine-tuned with custom dense layers, dropout tuning, and EarlyStopping.

→
04

Evaluation

Accuracy, precision, recall, F1-score, ROC/AUC, and confusion matrices across every architecture and split.

Benchmark across 8 architectures

Model Best Accuracy AUC F1 (macro)
ResNet50 99.41% 98% 93%
ResNet101 99.32% 97% 90%
MobileNetV2 98.97% 94% 91%
VGG19 98% 95% 89%
VGG16 98% 96% 96%
EfficientNetB0 98.68% 98% 91%
DenseNet121 96.79% 97% 86%
DenseNet201 97.62% 95% 82%

ResNet50 led individual models thanks to its residual connections, which support deeper learning without gradient degradation. A hybrid EfficientNet + MobileNetV2 ensemble pushed further — 99% precision, 97% recall, 96–97% F1 across both classes.

STACK

Python · TensorFlow · Keras · OpenCV · NumPy · Pandas · Matplotlib · Scikit-learn · Jupyter · Google Colab

NEXT STEPS

Multi-class classification (pneumonia subtypes, bronchiolitis), Grad-CAM explainability for clinician trust, and deployment as a lightweight web or mobile triage tool.

04 — SKILLS

Two toolkits, one instinct

Programming

PythonMATLABC++Arduino

AI / Machine Learning

Deep LearningCNNsTensorFlowKerasTransfer LearningROC / F1 Evaluation

Biomedical Engineering

Medical ImagingX-ray AnalysisBiomedical DevicesCalibrationPreventive Maintenance

Tools

JupyterOpenCVNumPyPandasScikit-learnGoogle Colab
05 — CONTACT

Let's talk.

Open to roles in biomedical equipment engineering, clinical support, and AI for medical imaging — and to research collaborations along the way.

EMAIL daniabeloufa22@gmail.com PHONE +213 673 327 251 LINKEDIN linkedin.com/in/dania-beloufa GITHUB github.com/daniabel90

This opens your email client with the message pre-filled, addressed to Dania directly.