Sitemap
A list of all the posts and pages found on the site. For you robots out there, there is an XML version available for digesting as well.
Pages
Posts
MooreSpeechCorpora Toolkit, Collecting Mooré Data from the Bible
Published:
Working with under-resourced languages like Mooré is both exciting and challenging. In this tutorial, I’ll walk you through the full pipeline I used to collect and prepare Mooré speech data scraped from the Bible-ready to be pushed to the Hugging Face Hub.
MooreSpeechCorpora Toolkit, Collecte de données Mooré à partir de la Bible
Published:
Travailler avec des langues sous-dotées comme le mooré est à la fois passionnant et exigeant. Dans ce tutoriel, je vous présente le pipeline complet que j’ai utilisé pour collecter et préparer des données de parole mooré extraites de la Bible, prêtes à être publiées sur le Hugging Face Hub.
YOLO, You Only Look Once
Published:
Abstract
We present YOLO, a new approach to object detection. Prior work on object detection repurposes classifiers to perform detection. Instead, we frame object detection as a regression problem to spatially separated bounding boxes and associated class probabilities. A single neural network predicts bounding boxes and class probabilities directly from full images in one evaluation. Since the whole detection pipeline is a single network, it can be optimized end-to-end directly on detection performance. Our unified architecture is extremely fast. Our base YOLO model processes images in real-time at 45 frames per second. A smaller version of the network, Fast YOLO, processes an astounding 155 frames per second while still achieving double the mAP of other real-time detectors. Compared to state-of-the-art detection systems, YOLO makes more localization errors but is less likely to predict false positives on background. Finally, YOLO learns very general representations of objects. It outperforms other detection methods, including DPM and R-CNN, when generalizing from natural images to other domains like artwork.
YOLO, You Only Look Once
Published:
Résumé
Nous présentons YOLO, une nouvelle approche de la détection d’objets. Les travaux antérieurs réutilisent des classifieurs pour la détection. Nous formulons plutôt la détection comme un problème de régression : boîtes englobantes et probabilités de classe en une seule évaluation. Un seul réseau prédit boîtes et classes directement à partir de l’image entière. Comme tout le pipeline est un seul réseau, il peut être optimisé de bout en bout.
ResNet, Deep Residual Learning for Image Recognition
Published:
Abstract
Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions.
ResNet, apprentissage résiduel profond pour la reconnaissance d’images
Published:
Résumé
Les réseaux de neurones plus profonds sont plus difficiles à entraîner. Nous présentons un cadre d’apprentissage résiduel pour faciliter l’entraînement de réseaux nettement plus profonds qu’auparavant. Nous reformulons explicitement les couches comme l’apprentissage de fonctions résiduelles par rapport aux entrées, au lieu de fonctions non référencées.
awards
Head of State’s Award of Excellence Permalink
Published:
Ranked 4th nationwide out of 154,775 candidates in the Baccalaureate.
Kingdom of Morocco Cooperation Scholarship Permalink
Published:
Merit-based scholarship covering 5 years of engineering studies.
Winner - Grand Débat d’Oujda Permalink
Published:
National debate competition winner showcasing teamwork and analytical skills.
2× AEBM Awards of Excellence Permalink
Published:
Recognized for academic excellence and community contributions.
2nd Place - Pan-African Data Science Nations Cup Permalink
Published:
Team Lead - Competed against 100+ teams across Africa.
Mitacs Globalink Research Award Permalink
Published:
International research collaboration award for Quantum ML research.
certifications
Machine Learning Specialization
Published:
Core ML concepts: supervised, unsupervised, neural networks and best practices taught by Andrew Ng.
Machine Learning Engineer
Published:
MLOps, MLflow, Docker, CI/CD, ETL, data versioning and monitoring for production ML systems.
Associate Data Scientist
Published:
Track covering Python, SQL, statistics, probability, data manipulation & visualization and predictive modeling.
Microsoft Certified: Azure Data Scientist Associate
Published:
Certified expertise in building, deploying, and monitoring ML models on Azure using Python and MLflow.
Oxford Machine Learning Summer School (OxML)
Published:
Theoretical foundations of ML from statistical methods and optimisation to representation learning and GenAI.
Data Engineering Specialization
Published:
Data engineering lifecycle: ingesting, storing, transforming, and serving data on AWS cloud.
extra
portfolio
Generative Adversarial Networks (GANs)
Published:
Deep Convolutional GAN implementation in PyTorch for generating realistic handwritten digits from the MNIST dataset.
Technologies: PyTorch, Deep Learning, Generative Models
Radio Signals Classification
Published:
Deep learning model for classifying radio signals using spectrogram images with PyTorch and transfer learning.
Technologies: PyTorch, timm, Audio Processing, Transfer Learning
Florence-2 Vision Language Model Toolkit
Published:
Comprehensive collection of notebooks for Microsoft’s Florence-2 VLM: inference, data auto-labeling, and fine-tuning for various computer vision tasks.
Technologies: PyTorch, Hugging Face Transformers, Computer Vision
Fine-Tuning MMS Adapter Models for Low-Resource ASR
Published:
Scripts and utilities for fine-tuning Meta’s Massive Multilingual Speech (MMS) adapter models for ASR on low-resource languages like Mooré.
Technologies: PyTorch, Hugging Face Transformers, Speech Processing
Energy Demand Forecasting for Africa - DataTour 2024
Published:
2nd place solution at the pan-African Data Science Nations Cup. Developed regression models to predict energy demand across 50+ African countries.
Technologies: Scikit-learn, Pandas, NumPy, Data Visualization
WrinnnAI - AI Mobile Assistant for the Visually Impaired
Published:
Real-time streaming mobile assistant using Computer Vision and Speech processing to help visually impaired users navigate their environment.
Technologies: Python, Computer Vision, VLMs, TTS, Flutter, FastAPI
RNA-Seq Pan-Cancer Clustering for Tumor Identification
Published:
Unsupervised clustering of gene expression data using K-Means and DBSCAN for automatic tumor type identification from TCGA dataset.
Technologies: Python, Scikit-learn, PCA, t-SNE
Tagantino AI - Smart Agriculture with Satellite Data
Published:
ETL pipelines and Deep Learning models to forecast droughts from satellite data, helping Moroccan farmers optimize water resources.
Technologies: PyTorch, AWS Glue, Airflow, LangChain, RAG
Moore Speech Corpora Toolkit
Published:
Open-source toolkit to collect, preprocess, align, and normalize Mooré language speech/text data for low-resource NLP applications.
Technologies: Python, Speech Processing, TTS, ASR
Spotify Batch Data Processing
Published:
Python project for extracting and processing data from Spotify Web API with OAuth 2.0 authentication and pagination strategies.
Technologies: Python, Spotify API, OAuth 2.0, REST APIs
Barbados Traffic Analysis Challenge
Published:
ML solution for predicting traffic congestion using video analysis. Implemented YOLO detection, VideoMAE embeddings, and ensemble models.
Technologies: PyTorch, YOLOv11, VideoMAE, XGBoost, OpenCV
publications
talks
Markov Chains and AI Applications
Published:
Presented Markov Chains and their applications in AI: weather forecasting, text and image generation.
Journée d’Orientation aux Métiers du Numérique
Published:
Panel discussion guiding future African talents towards careers in Data Science and AI.
LLMs: Opportunités pour l’Afrique
Published:
Traduction ? Culture ? Innovation ? Quelles sont les opportunités des Large Language Models pour l’Afrique ?
Préparer sa Carrière à l’ère de l’IA
Published:
Comprehensive webinar on AI opportunities, threats, and how to build a resilient career in the AI era.
teaching
Data Science & AI Summer Bootcamp
Intensive Bootcamp, GO AI Corporation, 2024
Led a 3-week intensive Data Science & AI bootcamp for 120+ learners at GO AI Corporation.
Python Programming Classes
Community Course, Community Initiative, 2025
Delivered weekly Python programming classes to 60+ students through a community initiative from March to July 2025.