Mohamed Seyam

Artificial Intelligence Engineer

Specialized in Computer Vision, Deep Learning, and MLOps. Building intelligent systems for medical imaging, satellite analytics, and automated workflows. Passionate about pushing the boundaries of AI to solve real-world problems.

mohamedseyam099@gmail.com
(+20) 01090982078
Cairo, Egypt
LinkedIn Profile GitHub Portfolio

Professional Experience

Artificial Intelligence Engineer

Wakeb Data · Cairo, Egypt

Jan 2025 - Present
  • Developed end-to-end AI systems for computer vision and geospatial analytics, including satellite image segmentation, object detection, and change detection
  • Led the annotation team, gathering project requirements, delegating tasks, and reviewing work for accuracy and consistency
  • Fine-tuned open-source Vision-Language Models and foundational models for land cover mapping and environmental monitoring tasks
  • Optimized deep learning models for deployment on edge devices such as NVIDIA Jetson, and containerized workflows using Docker
  • Collaborated with GIS and web teams to integrate AI solutions into production pipelines
  • Contributed to the development of an agentic RAG system for Robotic Process Automation (RPA) workflows
  • Played a key role in developing an agentic RAG solution for automated presentation generation
TensorFlow PyTorch Docker NVIDIA Jetson Open Vocabulary MLflow Qdrant pgvector GitHub Actions Nginx Ollama

Artificial Intelligence Engineer

Medsoft · Remote

Jan 2023 - Jan 2025
  • Built advanced 3D segmentation models in TensorFlow to enhance accuracy in medical imaging
  • Utilized AWS SageMaker with S3 storage to train and test large-scale 3D models effectively
  • Applied Statistical Shape Modeling combined with deep autoencoders for generative reconstruction and synthetic data creation
  • Implemented an innovative 3D landmarking approach using a 2D U-Net model, integrating computer graphics with deep learning
  • Collaborated closely with international teams, including the DePuy Synthes R&D team at Johnson & Johnson
TensorFlow PyTorch AWS SageMaker S3 U-Net GANs Autoencoder 3D Segmentation

Computer Vision Engineer

DevisionX · Remote

Nov 2022 - Jan 2023
  • Developed an automated machine learning pipeline for image classification, object detection, and segmentation using TensorFlow AutoML
  • Streamlined model training and deployment processes
AutoML Keras-tuner YOLO TensorFlow PyTorch

Featured Projects

Automatic Identification of Swallowing Kinematics in VFSS

View Project →

A deep learning toolkit designed to identify swallowing kinematics automatically in Videofluoroscopic Swallowing Study (VFSS). Incorporates advanced techniques including optical flow, action recognition, object tracking, and image segmentation to develop a computer-aided system that assists doctors in diagnosing dysphagia.

Medical Imaging Deep Learning U-Net SSD Optical Flow PyTorch

Needle - Deep Learning Framework

View Project →

A deep learning framework developed from scratch, built on the concept of automatic differentiation similar to PyTorch. Implemented using Python and C++ and serves as a foundational building block for constructing complex neural networks.

Automatic Differentiation Tensor Python C++ OOP

mini-RAG

View Project →

Developed a production-ready Retrieval-Augmented Generation (RAG) application as an educational project. Utilized Python with FastAPI framework and integrated with various databases. Deployed using Docker and Docker Compose with monitoring via Prometheus and Grafana.

Docker FastAPI PostgreSQL pgvector MongoDB Prometheus Grafana Nginx

Skills & Technologies

Programming Languages

Python C++ Scala

ML/DL & MLOps

TensorFlow PyTorch Hugging Face Transformers GANs VAEs Vision-Language Models Statistical Shape Modeling

Computer Vision

Satellite Image Segmentation Object Detection Change Detection Action Recognition Object Tracking Image Segmentation

DevOps & Build Systems

Docker Docker Compose Prometheus Grafana Nginx GitHub Actions Alembic Make

Cloud Computing

AWS S3 AWS SageMaker

Databases & Web Development

MongoDB PostgreSQL pgvector Qdrant Flask FastAPI RESTful API

Computer Graphics

Libigl Scalismo VTK CGAL

Other Tools

NVIDIA Jetson Ollama LaTeX Jupyter VSCode

Blog Posts

احدث نماذج نفيديا

Read Article →

NVIDIA, the company you might associate more with graphics and gaming, has just made a bold move into the world of artificial intelligence with the release of its Llama 3.1-70B Instruct model. This model is open-source, incredibly powerful, and directly competing with industry heavyweights like GPT-4.

NVIDIA Llama 3.1 Open Source AI LLM Arabic

هل الذكاء الاصطناعي بيفكر ولا بيحفظ ؟

Read Article →

Is AI Really Thinking? Apple's Research Exposes Alarming Flaws in AI Decision-Making. Apple's new research reveals that AI systems, even the most advanced, might not be truly thinking at all. Instead, they could be dangerously vulnerable to small, seemingly insignificant changes.

Apple Research AI Reasoning GSM-Symbolic Pattern Matching Arabic

Publications

AI-Powered Toolkit for Automated Swallowing Kinematic Analysis

AI-Powered Toolkit for Automated Swallowing Kinematic Analysis in X-Ray Videofluoroscopy

2022

Developed an AI-powered toolkit for automated identification of swallowing kinematics in Videofluoroscopic Swallowing Study (VFSS). The system uses advanced deep learning techniques including optical flow, action recognition, object tracking, and image segmentation to assist doctors in diagnosing dysphagia. The pipeline reduces initial examination time by over 50% and provides objective measurements for hyoid bone displacement and bolus tracking.

Medical Imaging Deep Learning Computer Vision Healthcare AI

Education

Cairo University

Faculty of Engineering

Bachelor's in Biomedical Engineering

2017 - 2022