About

I am an M.Sc. Informatics student at the Technical University of Munich, working on computer vision and 3D perception. My current research focuses on monocular and multi-modal 3D scene understanding, with an emphasis on geometric representation learning, segmentation, and vision foundation models for robust perception.

Before TUM, I worked as a machine-learning engineer at Aigorithm on computer-vision systems for object detection, segmentation, and active-learning data labeling. I also interned with the Zurich-based Google Shopping team, working from the Munich office on a machine-learning system prototype for offer freshness. These experiences shaped my interest in research that connects visual representation learning, 3D geometry, and reliable perception in real-world settings.

Research Interests

My interests center on robust scene understanding from visual and multi-modal data.

• Monocular and multi-modal 3D scene understanding for autonomous driving and robotics.
• Geometry-aware perception: 3D detection, segmentation, tracking, depth/shape estimation, and scene reconstruction.
• Vision and geometric foundation models: developing, adapting, and evaluating models for robust spatial perception.
• Weakly supervised, self-supervised, and scalable learning from images, masks, depth, and 3D structure.

Experience & Education

• M.Sc. Informatics

  Technical University of Munich

  May 2022 - Dec 2026 expected

  Expected graduation: Dec 2026. Focus: computer vision, machine learning, and 3D scene understanding.

• Research & Teaching Assistant

  Technical University of Munich

  Dec 2022 - Apr 2025

  Teaching assistant for Computer Vision 3: Detection, Segmentation and Tracking. Researched unsupervised semantic segmentation with denoising diffusion and student-teacher distillation.

• Software Engineer Intern

  Google Shopping, Zurich team; based in Munich

  Aug 2022 - Dec 2022

  Developed a machine-learning system prototype for estimating offer freshness from product and price signals.

• Junior Machine Learning Engineer

  Aigorithm, Cairo

  Nov 2020 - Apr 2022

  Developed computer-vision pipelines for data labeling, active learning, object detection, segmentation and model deployment.

• Software Engineer Intern

  DevisionX, Cairo

  Jun 2018 - Aug 2018

  Worked on ID verification using segmentation and recognition of Arabic text in TensorFlow, with Python/Flask backend integration.

• B.Sc. Computer Science

  Thebes University

  Oct 2014 - Aug 2018

  Final GPA: 3.44/4.00. Thesis: Arabic Image Captioning.

Projects

Selected Research Projects

• M.Sc. Thesis / Current Research: Monocular and Multi-Modal 3D Scene Understanding PyTorch 3D Perception Autonomous Driving
  Researching foundation-model-based and geometry-aware approaches for 3D scene understanding in autonomous-driving scenarios, with emphasis on robust perception and reliable evaluation.
• Guided Research: Multi-Modal 3D Object Detection [report] LiDAR Image Fusion
  Designed a LiDAR-image 3D detector with relation-aware reasoning for driving scenes and studied feature fusion strategies for 3D box localization.
• IDP: Semi-Supervised Vehicle Part Segmentation [report | code] Point Clouds Segmentation
  Explored semi-supervised vehicle part segmentation in LiDAR point clouds to reduce annotation cost for autonomous-driving perception.
• Master-Praktikum: Unsupervised Video Segmentation [report | code] Video Optical Flow
  Combined motion and appearance cues for unsupervised video object segmentation with optical-flow-based consistency losses.
• 3DMM Estimation from Highly Distorted Images [code | report] 3D Reconstruction
  Estimated 3D morphable model parameters from highly distorted images and analyzed robustness under camera distortion.
• Semantic Segmentation with DGCNN on ScanNet [code | report] DGCNN ScanNet
  Implemented point-cloud semantic segmentation on the ScanNet indoor dataset.
• Cobb Angle Estimation for Scoliosis Screening [project | hackathon] Medical Imaging
  Built a deep-learning pipeline to estimate the Cobb angle from spine X-ray images for automatic scoliosis screening.

Additional Technical Projects

• Road Damage Detection PyTorch YOLOv5
  Fine-tuned object-detection models and adapted training code for custom road-damage datasets.
• Human Action Recognition Keras Video
  Built video-classification models for human action recognition on UCF101.
• Arabic Image Captioning Keras NLP
  Prototyped Arabic image captioning using translated Flickr8k captions and Arabic word embeddings.
• Predict Future Sales scikit-learn Time Series
  Solved a Kaggle forecasting task with feature engineering, lag features, mean encoding, and ensemble models.
• Gender-Age Prediction Keras Faces
  Built a face-attribute prediction prototype using pretrained convolutional networks.
• Web and C++ Systems [Q&A | store | courses | tic-tac-toe] JSP MySQL C++
  Built full-stack, backend, and algorithmic software prototypes during undergraduate and self-directed work.

Awards & Competitive Programming

• ICPC: Qualified for the ICPC World Finals through ACPC, with strong regional results including 6th place at ECPC and 2nd place at ECPCQ.
• Meta Hacker Cup: Reached Round 2 in 2019 and 2021; among the top contestants in Egypt.
• Contest judge: Served as ICPC contest judge at several Egypt local contests and regionals.
• Online certificates: Udacity Machine Learning Engineer Nanodegree scholarship; completed in 3 months vs. 6 expected. Also completed the Deep Learning Specialization by deeplearning.ai.