- Audio2Audio DL Models, for guitar effect application- Analysis and ETL pipeline of the product usage data- Audio2Audio encoder-decoder architectures research

- Guitar Tone Classification- Guitar Tone Embeddings (DML)- Data Generation- User Data Analysis

Estimating car damage costs after accidents from images:- Data collection – Data Labeling (Accidented car images)- Deep CNN to identify the damages from an image- Postprocessing the output and converting to costsProject report: github.com/KaisKermani/car-damage-api

Summary: Predicting market futures price movement in real-time .Description:This project required a broad range of skills, it included:- Extracting historical market data using the the interactive brokers API- Analyzing and processing the historical data in order to build an accurate AI Model- Tuning and Testing the performance of different machine learning models.- Predicting the next movement of the price based on the actual and previous bid and ask activities

This project's aims to make carpooling more accessible to usual Tunisian Carpoolers.We built a software application that collects data from various Facebook groups (dedicated to carpooling), parses it using and NLP model, and then stores it and displays it in a structured way on a web interface.The whole work required expertise in preprocessing and post processing, as well as knowledge in Named entity recognition and CRF.

Developed a desktop application that generates and exports relevant graphs from csv files.

I worked on various embedded and IOT projects that involve working with Esp, ATMega and STM µControllers.