Hi, I'm
Mohamed Ouchraa
Machine Learning & AI Engineer
I'm specialist in
Hi, I'm
Machine Learning & AI Engineer
I'm specialist in
Hello! I'm Mohamed Ouchraa, a passionate Machine Learning and AI Engineer.
I am a Master's student specializing in Advanced Machine Learning and Multimedia Intelligence, with a strong passion for artificial intelligence, computer vision, and ai generative technologies .
My academic and project experience has equipped me with solid skills in machine learning, data science, software development, and complex problem solving. I bring a rigorous approach to data analysis combined with the ability to design innovative, practical solutions. Curious, proactive, and a strong team player, I thrive in collaborative environments where I can turn ideas into real-world impact.
Creating novel AI solutions
Deep expertise in ML techniques
Working effectively in teams
The system receives user emails, uses a LLM enhanced with Retrieval-Augmented Generation (RAG) to generate SQL queries, and produces reports to automate analytics requests in the telecom domain.
Cleaning, preparation, and analysis of large-scale Covid data. Implementation of ML models (Random Forest) and DL (LSTM) for case prediction.
Twitter data collection via web scraping, preprocessing with NLTK, and development of a sentiment analysis model. Creation of an interactive dashboard.
Development of a web application with Django to manage and visualize geospatial data (OSM). Searching for nearby points of interest.
Design of a graphical interface in Matlab to apply filters (high-pass/low-pass) and Hough transformation for shape detection.
Loading, visualization, and training of a PointNet model on the ModelNet10 dataset for 3D object classification from point clouds.
Django web application using SQLite to store user information and implementing facial recognition algorithms for authentication.
Preprocessing of spatiotemporal data and training of a GRCN model to predict collision risks. Interpretation with SHAP.
Using YOLOv5 and Roboflow to train an object detection model (players, ball, referee) in tennis matches. Real-time prediction.
Segmentation of satellite images to detect flood-affected regions using Reinforcement Learning. Agents trained with A2C, PPO, and DQN adaptively optimize segmentation masks.
Detection of pedestrians using YOLOv5 on real-time surveillance data. Explainability with SHAP, Grad-CAM, and LIME to interpret model decisions and highlight important regions.
A Django and SQLite-based web platform for managing PhD thesis recruitment. Doctoral students can apply and select thesis topics. Professors propose topics and select candidates. Admins monitor activities through a dashboard, and the Dean validates the final thesis selections.
Real-time detection of Moroccan vehicle license plates using YOLOv5. The model is trained on a custom dataset with local plate formats. Application includes video stream processing and bounding box annotations.