Merwan Amar Chelouah

Within the AI Center of Excellence at Sanofi, I design and operate the data platforms and pipelines that power ML models and AI products in production for the global manufacturing and supply chain.

Design of end-to-end data pipelines for AI/ML use cases: feature engineering, building training datasets, and inference pipelines.

Provisioning reliable, high-quality data throughout the model lifecycle.

Close collaboration with Data Scientists on model development, production deployment, and monitoring.

Optimization of data systems for large-scale, real-time processing on AWS.

Implementation of CI/CD and automation for data and ML pipelines (MLOps).

Ensuring data quality, governance, and reproducibility across ML workflows.

Contribution to robust and scalable data foundations in a global health context.

Stack: Python, SQL, Spark, Snowflake, Airflow, dbt, Terraform, Docker

Leading the design and implementation of cloud-native, event-driven data architectures to support scalable and reliable organization-wide analytical solutions.

Key responsibilities and achievements:

Design and development of data pipelines on AWS, using Lambda, Step Functions, EventBridge, SQS, and SNS, to enable seamless ingestion, transformation, and delivery of data in various formats (CSV, JSON, Parquet) to S3, relational databases, REST APIs, and Snowflake.

Development and deployment of a self-service Datalab platform (React + Node.js), hosted on AWS (ECS, Route 53, ALB), enabling Data Scientists, Business Analysts, and business teams to provision isolated environments with governed access to data from the enterprise DataHub.

→ Drastic reduction in data provisioning time—from several weeks to a few hours—accelerating experimentation, prototyping, and time-to-insight within the organization.

Automation of infrastructure provisioning via AWS CDK (Infrastructure as Code) and implementation of robust CI/CD pipelines with Jenkins to industrialize and ensure reliable deployments.

Leading backlog prioritization for architecture and delivering end-to-end architecture studies and proof-of-concepts (POCs) for strategic data initiatives.

Writing detailed architecture blueprints, technical documentation, and high-level designs to guide implementations and ensure alignment with enterprise standards.

Hands-on contribution to key technical topics while supporting tech leads in defining and applying engineering best practices within the data platform.

Promotion of a data-first approach, with a strong emphasis on automation.

Leading end-to-end data integration processes, designing and orchestrating Credit du Nord data ingestion pipelines using Python, SQL, and ETL frameworks, ensuring data integrity and alignment with Franfinance's risk architecture.

Evaluation and mitigation of systemic impacts across the entire data value chain, including model drift, feature stability, and data lineage issues affecting regulatory models (PD, LGD, EAD).

Leading regression testing, backtesting, and stress testing campaigns to validate model performance post-integration, in compliance with Basel III/IV requirements and internal validation frameworks.

Building regulatory risk analysis frameworks to quantify capital and pricing impacts related to portfolio absorption, leveraging advanced metrics (RWA, capital buffers) to support pricing strategy and decision-making.

Automation of operational risk processes and reporting pipelines in a Unix environment (shell scripts), with downstream processing in SAS and Python, reducing execution time and manual intervention by over 60%.

Key interface role between Data Science, Risk, and IT teams to ensure data traceability, compliance with governance requirements, and preparation for model audits throughout the project lifecycle.