My Projects

For my PFE, at Capgemini Engineering,I have trained and fine-tuned Large Language Models (LLMs) and applied Retrieval-Augmented Generation (RAG) techniques to enhance performance. Additionally, I utilized AWS GenAI services to develop an interactive Generative AI-based assistant system. My work also involved designing an advanced multi-document data extraction pipeline and implementing feedback systems that incorporate the latest innovations in Generative AI.

Worked on creating a streaming app for streaming movies, animes and series with a mini Blog for movie reviews, Using SPRING and Angular with MongoDB, we applied webscraping and third party APIs to get external Data and play the movies on our app with multiple quality settings.

This mini project focuses on detecting documents in a video stream or image using computer vision techniques and applying a perspective transformation to obtain a flattened view of the document. The primary goal is to automate the process of identifying documents within a scene, even when the documents are tilted or skewed, and then rectify them to appear as if viewed from directly above.

Trained a CNN classifier using 'FER-2013 dataset' with Keras and tensorflow backened. The classifier sucessfully predicted the various types of emotions of human. And the highest accuracy obtained with the model was 60.1%. Then used Open-CV to detect the face in an image and then pass the face to the classifer to predict the emotion of a person.

Developing machine learning models for the automated detection of bird and frog species in tropical soundscapes by creating a system that can accurately identify species with limited training data.

Designing and developing a machine learning model, based on some data for transactions, in order to automate fraud detection, with the use of Neo4j graph database.

Used Text Mining with multiple techniques to create a model that can classify customer complaints, we traine the model using Decision tree algorithm and Random Forestn with features selection : Genetic Algos, Mutual Info, Info Gain. We managed after tuning the parameters to get a maximum accuray of 82%.

Designing and developing a desktop app to manage all the services inside a PC repairs and sales store, with an employee and storage managament system added to it.

A template of a wbsite for selling and buying both used and new cars using HTML, CSS and Js only with possibility to view cars informations and react to posts. Visit website for a better view.

A Mern stack application to manage and schedule tasks for Employees and teams with Notification System.

A simple Tic Tac Toe Game using react.js and socket.io where palyers can join a room and compete agaisnst each others.

Simple blog for a Hair dressing community to share posts and discuss some topics, built with PHP Laravel and mysql database following the MVC architecture and bootstrap for the front end.

A fully working desktop application designed and developed with python using different machine learning libraries that is used to detect students attendance in a classroom based on facial recognition.

Full .net App with database for renting out cars in Morocco using online application forms and the ability to manage the cars and applications.

Worked with a team on a school project to build a website for job seekers and recruiters in the IT domain with many features and functionnalities like the ability to create a resume using the built in feature or to apply directly to the jobs from the app.

A web application built with J2EE using Spring Boot Framework, in order to manage storage of different products and keep track of the transactions going on.