Abdelaziz Hamadi

R&D engineer specialized in computer vision, image processing, and machine learning (MSc), with a demonstrated history of work on my github. Skilled in various programming language and fields such as Deep Learning, visual odometry, image classification & object recognition. Also, solid knowledge in 3D geometry and computer graphics.

Pristine Robotics | Rennes, France

R&D Computer Vision Engineer & Consultant

• Proven expertise in translating the client’s specific needs into easy-to-use software. 
• Provided consulting services and writing process documentation for delivery reports.
• Market studies to help determine service pricing based on market trends and demand.

• Implemented surface defect segmentation using DL for products’ surface inspection. Achieved 73% accuracy on dice score.
• Implemented unsupervised machine learning algorithms for products’ absence counting during production. 
• Parallel computing and vectorization using CUDA and threading. Optimized real time performance by 40%.
• Hands-on experience in end-to-end data pipelines. From live video stream to a server-based monitoring system for visualizing time series data and triggering alerts if required.
• Developed a user-friendly interface to control different algorithm parameters and trigger real-time inspection on production sites.

R&D Computer Vision Engineer Intern

• Developed a recommendation for an inspection box that effectively adapts to the company’s production lines for product quality control.
• Conducted research on state-of-the-art object detection techniques in computer vision and develop a deep learning-based detector capable of identifying the company's baked goods.
• From data collection, to training and evaluating different Deep Learning models performances using different scientific metrics.
• Designed and implemented state-of-the-art computer vision solutions for products quality inspection, such as roundness, colour, and gliding of the products.

R&D Computer Vision Engineer Intern

• Conducted research on state-of-the-art on multi-object tracking algorithms. 
• Retrained existing object detection models on hand crafted CCTV camera dataset using transfer learning for person class only. 
• Bridging between object detector and DeepSORT algorithm for detection-based tracking. 
• Achieved better performs in terms of multi-person tracking especially in extremely low light conditions. 

Master 2 Signal and image processing: IMAGES, MODELS AND VISION | Sept/2022

Aix-Marseille University, Marseille, France

End of study project: " multi-person tracking and heat map establishment”. Grade: 15.17/20

Masters: Computer Science, Image & Artificial Life | Jul/2022

Mohamed Khider University, Biskra, Algeria

End of study project/internship: " multi-object tracking using Deep Learning in real-time”. Grade: 17.33/20

Bachelor of Science: Computer Information Systems | Jun/2019

Mohamed Khider University, Biskra, Algeria

End of study project: “E-learning Android application for programming languages”. Grade: 18/20