Big Data & Cloud Analytics

In projects with over a hundred dbt models on BigQuery, tests run into the thousands. Data that fails them never reaches the gold layer. Medallion architecture, from ingestion to the dashboard.

Let's talk data

“Data models tend to evolve more slowly than code, and the impacts of poor data models are sneaky. You won't know you're in danger until it's too late. So, take the time to do things right. Move slowly to move faster and safer over the long haul.”

— Joe Reis

A data platform has to survive years of business, schema changes and campaign peaks.

We build them the way serious software gets built. Every transformation versioned in git, every dataset tested, every flow wired to an alert.

Our strength is analytics, and we run dbt on both sides: in the warehouse and in the operational database.

dbt ≠ pipeline

dbt is the T in ELT, and we treat it the way good software gets treated, the pattern the industry calls analytics engineering. But the tool is only one piece: ingestion, orchestration and monitoring live elsewhere, and our value is building and running the whole set.

reliable_metrics.sql ✕▶ Run

-- which numbers do you trust?
SELECT metric, lineage, last_run
FROM `gold.business_metrics`
WHERE reliable = TRUE;
-- in one project: 1,646 tests across 111 dbt models

✓ Query complete (1.4 s) · every metric's lineage, from ingestion to analysis

Row	metric	lineage	last_run
1	daily_sales	complete	4 min ago
2	stock_by_warehouse	complete	4 min ago
3	margin_by_promo	complete	4 min ago

Data that doesn't add up never reaches gold. That's the governance.

Google Cloud is where we've gone deepest. BigQuery as the warehouse, dbt for the transformations.
From whiteboard to publication. Our definition of done starts long before the code. Database design following the sector's classic modeling, standard naming, homogenization and normalization of the data. It is more analysis than development, and it is where reliability is won.
Testing is the part most teams skip. On large projects, tests number in the thousands. Unit tests run before anything ships, and data tests run every day, with every load.
Layered architecture. The medallion pattern keeps raw and verified data apart, each layer with its purpose and its access control.
Infrastructure is code too. Terraform builds the same datasets, permissions and environments in development and in production.
Real time when the business needs it. Ingestion on Pub/Sub and Dataflow for the flows that can't wait for the nightly load.

dbt on BigQuery and on Oracle

On the analytical side we work with dbt Cloud on BigQuery. And when a transformation belongs in the operational layer, on a relational database like Oracle, we manage it with dbt Core and the adapter Oracle maintains (plus our own improvements, as obvious as they were necessary...).

The same tests and the same versioning on both sides, with no toll booths between teams. Data reaches the dashboard already proven.

We connect BigQuery to your e-commerce, to your ERP, and to the edge.

Let's talk, engineer to engineer.

Bring the problem as it is. You'll hear back from the person who'll write the code.

Tell us about your project