About Khadija
Bonjour 👋
J’aide les entreprises à concevoir, entraîner et déployer des solutions d’IA et de machine learning, avec une spécialisation en NLP, reconnaissance vocale et IA générative.
Data Scientist & ML Engineer, j’interviens sur des projets end-to-end : collecte et préparation des données, entraînement et fine-tuning de modèles, évaluation, industrialisation et mise en production.
J’ai travaillé sur des problématiques variées, notamment dans des contextes peu documentés et multilingues, ce qui m’a permis de développer une forte autonomie technique et une capacité d’adaptation rapide.
Domaines d’intervention
- NLP & IA générative : classification de textes, RAG, traduction automatique, transcription audio, fine-tuning de modèles Transformer (BERT, NLLB, Wav2Vec2, etc.)
- Machine Learning & Deep Learning : CNN, Transformers, modèles d’attention, classification, régression, séries temporelles
- Data Engineering & pipelines : collecte, annotation, preprocessing et structuration de données
- MLOps & déploiement : MLflow, Docker, AWS, Azure, mise en production et monitoring de modèles
Stack technique
Python · PyTorch · TensorFlow · Hugging Face · Scikit-Learn · MLflow · Docker · AWS · Azure · PySpark · SQL
French
Fluent
English
Fluent
Arabic
Native or bilingual
Remote only
Primarily works remotely
Experience
- SiliconeSignal TechnologiesData ScientistAugust 2022 - December 2025 (3 years and 4 months)Meknes, Morocco
- Fine-tuned Wav2Vec2 models for Moroccan Darija speech-to-text transcription on low-resource audio datasets
- Collected, annotated, and validated custom multilingual datasets for NLP and speech recognition tasks
- Built a proprietary Darija ↔ English parallel corpus (7,300+ sentence pairs) for NLLB-200 translation fine-tuning
- Designed and benchmarked Transformer and CNN-Attention architectures for sequential learning tasks
- Developed an automated sleep stage classification pipeline using physiological signals (heart rate + respiration) on the MESA dataset (~2,000 subjects)
- Built end-to-end data preprocessing and training pipelines for low-resource Arabic dialect NLP applications
- Conducted model evaluation, experimentation, and performance optimization for production-oriented ML workflows
- AEVAWEBNLP Research InternMay 2022 - July 2022 (2 months)Oujda, Morocco
- Developed a French Question-Answering system using CamemBERT and PyTorch
- Implemented NLP preprocessing pipelines including POS tagging, lemmatization, and text normalization
- Built and evaluated Transformer-based NLP models on Wikipedia-based QA tasks
- INRAEData Engineering InternMay 2021 - October 2021 (5 months)Aix-en-Provence, France• • Automated an end-to-end data collection pipeline for flood event monitoring across France, aggregating and structuring heterogeneous data sources
Recommendations
Be the first to recommend Khadija
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Master – Data Exploration & Business Intelligence (EID²)Galileo Institute, Sorbonne Paris Nord University2021Master – Data Exploration & Business Intelligence (EID²)
- Bachelor of ScienceMoulay Ismail University2019Bachelor of Science