Focus
Real-time detection, medical imaging, visual analysis, and production ML services.
Saad Abd El-Ghaffar
Kaggle Expert · Computer Vision & Machine Learning Engineer
B.Sc. Computer Engineering · Mansoura University | Software Engineer @ TJM Labs
I build and deploy computer vision and ML systems — real-time object detection, medical image analysis, and visual pipelines — mainly with Python, PyTorch, and OpenCV.
At TJM Labs I work on AI agents and APIs that automate pharmacy workflows such as prescription intake and fax processing. Outside work I take projects from training to deployed services with FastAPI, Flask, and Docker.
About
I graduated with a B.Sc. in Computer Engineering from Mansoura University (94.08%, ranked 6th of 255, Excellent with Honors). My strongest work is in detection, medical imaging, and deploying models behind APIs.
At TJM Labs I build AI agents and backend services for US pharmacy systems — prescription intake, fax processing, and demographic updates. My graduation project, MathEditor, included OCR for math expressions to LaTeX and LLM-assisted document tools. I am preparing for a research Master’s in 2027.
Research
Mentored research training and collaborative paper work in computer vision and AI for healthcare.
Research Training Mentorship
5 monthsUnder Dr. Mohamed Sameer Abdallah, PhD, PMP® · Assistant Professor, Gachon University, South Korea
Completed a 5-month research training period focused on how to read papers carefully, assist in research discussions, and practice the process of writing research papers in computer vision.
What I Worked On
- Trained systematically on research-paper reading: problem setup, method, experiments, and how claims are supported by evidence.
- Read and studied ~30 research papers in the sign language recognition domain (vision-based gesture/sign understanding, related deep learning approaches, and evaluation practices).
- Assisted with research-oriented tasks and discussions under mentorship, including summarizing papers and connecting methods across the literature.
- Practiced academic writing fundamentals: structuring related work, clarifying contributions, and organizing experiment narratives.
Collaborative Research Paper — Under Review
Under reviewWith Dr. Tahir Abbas Khan (SMIEEE) · Assistant Professor, Muhammad Nawaz Sharif University of Engineering & Technology (MNSUET), Multan
Title: A Hierarchical Agentic Framework for Cardiovascular Triage: Design, Evaluation, and the Discovery of a Confidence–Severity Escalation Gap
Collaborative manuscript on an agentic, multi-stage cardiovascular triage pipeline (tabular risk screening → echocardiographic EF confirmation → MRI escalation), with confidence-based routing, ablations, and explainability/audit analysis. Status: under review.
My Involvement
- Contributed to the experimental pipeline and evaluation stages (screening, confirmation, escalation, orchestration, ablations, and explainability workflows).
- Supported manuscript development around methods, results, figures, and verification of reported claims against experiment outputs.
- Worked in an AI-for-healthcare research setting spanning medical imaging, tabular risk models, and agentic clinical decision pathways.
Experience
Software Engineer
Nov 2025 – PresentBuilding AI agents and backend APIs for US pharmacy workflows — prescription intake, fax processing, and demographic updates — wired into existing production systems.
AI Agents & Automation
- Automating pharmacy workflows that used to be mostly manual.
- Working with pharmacists and engineers on what actually needs to ship.
Backend Integration
- Developing services and APIs that plug AI into existing pharmacy software.
- Keeping deployments reliable enough for day-to-day operations.
Projects
Selected CV and ML projects with training, evaluation, and deployable demos.
Tennis Analysis System
Advanced match analysis using YOLO, deep learning, and real-time video analytics: player detection, ball tracking, court keypoints, and annotated performance statistics.
GitHub ↗Brain Tumor Classification
Fine-tuned ResNet18 for MRI classification with FastAPI backend, Streamlit frontend, and Docker packaging. Reported 99.7% accuracy and earned a Kaggle gold medal.
GitHub ↗License Plate Recognition
AI-powered plate detection with YOLOv11 and PaddleOCR, including a Flask interface and MySQL persistence for reads from video streams.
GitHub ↗Human Activity Recognition
Real-time activity recognition that classifies sitting and standing poses with YOLOv8 pose estimation and OpenCV video processing.
GitHub ↗Iranian Snack Detection
Smart retail detection pipeline with YOLOv11 for snack recognition, plus real-time price, expiry, and calorie analysis for automation workflows.
GitHub ↗Cats vs Dogs Classifier
TensorFlow / VGG16 image classifier at 98.7% accuracy, with FastAPI, Streamlit, Docker packaging, and a Kaggle gold-medal notebook.
GitHub ↗
Anime Face GAN Generator
DCGAN and CGAN models for anime face generation, with an interactive Streamlit app, controllable attributes, batch generation, and training notebooks.
GitHub ↗Hotel Cancellation Predictor
Reservation cancellation predictor at 97.2% accuracy, shipped with FastAPI, Streamlit, and Docker for revenue-oriented decision support.
GitHub ↗
Customer Segmentation (RFM)
RFM analysis and K-means segmentation with an interactive Plotly Dash dashboard for turning e-commerce data into actionable customer groups.
GitHub ↗
Ice Cream Sales Forecasting
Compared ARIMA, SARIMA, XGBoost, LSTM, and GRU on monthly sales data with preprocessing, feature engineering, and model performance analysis.
GitHub ↗Laptop Price Predictor
End-to-end price prediction with Random Forest, XGBoost, and CatBoost ensembles, feature engineering, and evaluation around ~91% R².
GitHub ↗Computer Vision Lab
Collection of realtime vision demos: gesture recognition, facial recognition, object detection, lane detection, and CNN image classification.
GitHub ↗ML From Scratch
Classical machine-learning algorithms implemented from first principles to keep theory tightly connected to code.
GitHub ↗Training / Certifications
Formal training programs and certificates from my CV.
WorldQuant University — Data Science Lab
Dec 2022 – Aug 2023Certificate
Completed 8 data science projects covering machine learning, time series analysis, and statistical modeling. Built predictive models with Linear Regression, ARMA, Logistic Regression, Random Forest, and GARCH, plus interactive dashboards with Python, Plotly Dash, and custom APIs.
Kaggle Expert
KaggleRank · Competition Medals
Kaggle Expert with multiple Gold medals (Cats vs Dogs, Brain Tumor Classification, Time Series Ice Cream Sales) and Silver medals (Customer Segmentation, Hotel Cancellation Predictor, Laptop Price Predictor).
Information Technology Institute (ITI) — Computer Vision Training
Jul 2024 – Sep 2024Certificate
Completed 5 advanced CV projects including tampering detection, image watermarking, text extraction, face swapping, and handwritten digit recognition. Used TensorFlow and PyTorch with data augmentation, feature extraction, and transfer learning.
Huawei — HCIA-AI Learning Course
Sep 2024Certificate
Hands-on experience with AI platforms focused on data processing, model training, and evaluation. Designed AI-based solutions using neural networks and optimization techniques.
Education
Bachelor's Degree
B.Sc. in Computer Engineering
Mansoura University, Egypt
Graduated
Excellent with Honors
94.08% · Ranked 6th / 255
≈ 3.8 / 4.00 equivalent
Contact
Reach me on Kaggle, LinkedIn, GitHub, or email.