Wicem Mokni

===== Context ====

Strategic migration from legacy SAP BW to Snowflake and Power BI. Enterprise scope: multi-brand (L'Occitane, Erborian, Melvita), multi-country (Europe, Asia, America). Integrated into a structured data team, with full ownership of assigned tasks, coordinating with relevant internal teams (IT, QA, Retail, E-commerce, Master Data).

==== Achievements ====

- Unified Customer Master Data: designed the multi-source ingestion architecture (POS, e-commerce, CRM, MDM) into Snowflake. Implemented loading pipelines, dbt transformation views, and data freshness monitoring via Airflow.

- Snowflake Data Sharing: cross-account access to customer master data with automated backup mechanism.

- Multi-country E-commerce Pipeline: incremental dbt models covering several international markets (UK, Japan, Australia, Ireland). Managed data temporal consistency constraints.

- Shopify Integration via Fivetran: connector configuration, dbt modeling of e-commerce data, and synchronization into the Snowflake data warehouse.

- B2C E-reporting Tax Compliance: designed the aggregation and export pipeline for tax declaration. Managed initial and corrective versions, Airflow orchestration.

- Quality & CI/CD: implemented a complete quality chain: dbt tests, automatic SQL linting, CI/CD pipeline on Bitbucket to ensure no regression in production.

==== Results ====

- Operational enterprise architecture: 3 brands, 15+ countries, 30+ data sources unified in a centralized customer master data repository.

- 60% reduction in e-commerce report delivery time thanks to incremental dbt models.

- B2C e-reporting tax compliance operational before regulatory deadline.

- dbt CI/CD deployed in production: zero regressions since implementation.

===== Context ====

Photovoltaic scale-up with international growth, no data team upon arrival; data was maintained poorly by back-end developers.

Mission: industrialize the data platform end-to-end (infra, DataOps, orchestration, modeling, dataviz) and support team upskilling on the modern data stack and production best practices.

==== Achievements ====

- Target Architecture: mapped existing systems and designed a cloud-native Snowflake + dbt + Airflow platform.

- Snowflake Security: full audit, access cleanup, MFA, RSA, role separation, and account-level network policy.

- Orchestration Overhaul: migrated ~35 jobs from an unreliable stack to containerized Airflow, with operational ETL → dbt dependencies and alerting.

- Airflow Industrialization: self-hosted deployment, git-sync, custom dbt image, GitLab CI/CD.

- Secrets Management: implemented AWS Secrets Manager with IAM, migrated credentials out of .env files.

- dbt Industrialization: DEV / STAGING / PROD environments, multi-stage CI/CD (SQLFluff, dbt compile, dbt test), freshness monitoring.

- Unified Customer Master Data: multi-source dimensional modeling (CRM, ERP, field, accounting) with reconciliation by similarity matching.

- Business Reporting: sales, profitability, after-sales service, and quality marts connected to Power BI for management and Metabase for operations.

- Team Acculturation: trained back-end developers on Airflow, dbt, Snowflake, and industrialization practices (Git, CI/CD, secrets, tests).

==== Results ====

- Industrialized data platform delivered in 3 months.

- +30% potential revenue identified.

- Prospect quality x2: 33,000 duplicates/triplicates cleaned.

- Reduced Snowflake attack surface.

- Daily installation time tracked by management thanks to Power BI dashboard.

- Team autonomous on the modern data stack.

===== Context ====

Sole Data Engineer for a CE-certified medical application. Infrastructure migration and implementation of an NLP pipeline for extracting clinical biological data (HDS/GDPR requirements).

==== Achievements ====

- Complete migration of the Claranet stack to a modern data stack on GCP / S3NS (data sovereignty, full autonomy).

- Complete NLP pipeline: event-driven ingestion (GCS + Pub/Sub), Mistral OCR (Vertex AI), extraction of biomedical entities (biomarkers, values, units, dates), semantic normalization.

- Structuring and storage of clinical data in BigQuery, transformation and data quality with dbt.

- Airflow orchestration (Cloud Composer), cloud-native pipeline deployed on Cloud Run (Git, logs, retries, monitoring).

- GDPR/HDS Compliance: anonymization, traceability, granular access control.

==== Results ====

- 85% reduction in manual processing time for clinical documents.

- 94% NLP extraction accuracy.

- Hosting costs reduced by 3x thanks to GCP migration.

- HDS compliance achieved in 2 months.

==== Key Skills ====

Python, Mistral OCR, Vertex AI, GCP (Pub/Sub, GCS, BigQuery, Cloud Run, Composer), dbt, Airflow, Terraform, GitHub Actions