Freelance Data & AI · France · Remote
Data Scientist and ML/AI Engineer. I build predictive models and RAG systems, fine-tune LLMs, and train teams to put AI to work.
open ./cv/mouaad-agourram --role data-scientist
Book a call
Freelance, available for missions in France or fully remote.
Daily toolkit
From framing a use case to evaluating the model, I cover the full ML lifecycle. Then I teach your team to run it without me.
Most of my work starts with a business question: what can we predict, and is it worth it? I design the experiments, train the models, and measure whether they deliver.
{
"name": "Mouaad AGOURRAM",
"location": "France",
"role": "Data Scientist / ML Engineer",
"focus": ["ML", "RAG", "LLM fine-tuning", "Forecasting"],
"availability": "Freelance / Remote / France"
}Three roles since 2024, one thread: take a model from first experiment to delivered result.
Design and delivery of workshops in Data Science, Deep Learning and GenAI. Participants leave with working code, not slides.
Forecasting and audio ML systems built end to end: data preparation, training, fine-tuning and evaluation against business metrics.
RAG pipelines and LLM fine-tuning prototypes, with evaluation datasets to decide what ships and what stays a demo.
Problem, approach, outcome — not just a category label.
A client needed automatic confirmation of purchases from audio streams. I built a deep learning model to detect the relevant keywords, then industrialized it on Azure Machine Learning — deployment, monitoring and drift tracking included, not left as a notebook.
Research collaboration on a self-adapting reinforcement-learning model for energy optimization that factors in user preferences. A working proof of concept, built to demonstrate the approach rather than ship as a product.
Training engagement covering advanced deep learning, Apache Spark, real-time data engineering and data visualization. Sessions built around working code, not slides.
Also on the list: quarterly sales forecasting with ARIMA/SARIMA, and a RAG assistant prototype.
A deliberately small toolbox: classic ML for tabular data, deep learning for the rest, and evaluation everywhere.
Tools I use to train, evaluate, fine-tune and explain useful AI systems.
$ cv skills --grouped
Machine learning:
scikit-learn, XGBoost, LightGBM, PyTorch, TensorFlow
Generative AI:
RAG, LangChain, FAISS, Hugging Face, LoRA / QLoRA
Forecasting:
time series, fine-tuning, evaluation, MLflow
Training:
Data Science, ML/AI, RAG, LLM fine-tuningFreelance engagements in France or remote, from a scoped prototype to a full training program.
Frame predictive use cases, build models, evaluate performance and make results understandable.
Prototype retrieval workflows, vector search, evaluation datasets and answer quality checks.
Prepare datasets, fine-tune models, compare baselines and evaluate task-specific performance.
Hands-on sessions where your team builds and evaluates real models on its own use cases.
Selected analyses published in full: methodology, code and results.
Binary classification of injury-accident severity on French road-safety data: baseline comparison, logistic regression and evaluation on imbalanced classes.
Lire le rapportSpatial and geostatistical analysis of income levels using INSEE's Filosofi gridded data (200 m cells): mapping a department split by inequality.
Lire le rapport