Tom Fournier

Within the Fuel, Gas & New Energies squad at E.Leclerc, I contribute to the valorization of data related to service stations in the network and energy initiatives.

My mission consists of identifying, structuring, and validating data pipelines to monitor the activities of E.Leclerc stations and Synergie partners, as well as indicators related to the energy performance of buildings and compliance with the tertiary decree.

I work on researching, analyzing, and integrating various data sources to build robust pipelines dedicated to:

- operational management of service stations,

- monitoring energy consumption and performance,

- and centralizing data necessary for managing regulatory obligations related to the tertiary decree.

This work improves the quality and availability of data for business teams, while facilitating energy performance monitoring and decision-making regarding fuel and energy transition issues.

Context:

I joined a large payment services company within the POS / Hardware POS tribe after a first failed dbt migration. The architecture was unstable, costly, and difficult to maintain, with several Snowflake projects far exceeding budgets.

Mission:

Restart the migration by starting almost from scratch, clean up the layers (staging, curated, mart), implement tests, consolidate existing projects, reduce Snowflake costs, and significantly improve execution performance.

Actions:

I have:

Completely revamped the dbt architecture and rebuilt the layers with clear conventions, clean lineage, and optimized models.

Implemented a comprehensive testing strategy (schema tests, relationships, data quality).

Migrated and consolidated POS Analytics (€1,500–€2,000/month), Tiller (~€2,000/month), and Online Store (~€350–€500/month).

Decommissioned costly legacy pipelines.

Optimized incremental models processing ~30 million transactions/day + millions of events.

Significantly improved performance: execution time reduced from approximately 2 hours to ~1 hour through structural optimizations and better volume management.

Stabilized Snowflake consumption (peaks ~€4,000/month) and unified all workloads into a single coherent project.

Documented and shared assets via dbt-loom.

Results:

Successful migration and fully cleaned dbt architecture.

Reduced Snowflake costs by 40–45% (stabilized around €2,500/month).

Doubled performance: executions went from ~2 hours to ~1 hour.

Scalable, reliable, maintainable platform, absorbing massive volumes at no extra cost.

Improved quality and transparency thanks to tests and clean lineage.

As a freelance consultant, I led several BI projects as a project manager and data solution architect. My role included creating data warehouses with dbt, data modeling, and setting up data ingestion pipelines. I also ensured data governance, implemented consistency tests, and provided structured data for data visualization solutions like Power BI.

Major Projects:

Iris Galerie (global network of iris photography boutiques)

Design and structuring of data for a global network of boutiques, managing different currencies and time zones.

Specific management of boutiques on cruise ships, where I handled multi-currency payments and VAT rules varying by ship location (international waters, home port).

Cardinet (medical imaging center in Île-de-France)

Creation of performance indicators to measure the activity of medical practices and doctors.

Development of a dashboard to track practice occupancy rates to optimize doctor schedules and improve shift efficiency.

Cabosse (Alain Ducasse – chocolate)

Implementation of stock monitoring for key events (Christmas, Valentine's Day, Easter), enabling production anticipation and management of product distribution across the boutique network, based on needs, losses, and sales during monitored events.

Lacoste (in-store sales promotion)

Implementation of performance indicators to track sales associate activity by store.

Development of a product recommendation logic, allowing sales associates to suggest high-performing items based on sales trends observed in other stores within the same network.