One trusted place for your company's data
A lakehouse and pipeline setup in your own cloud, with tested transformations, scheduled jobs, classified data and access rules that match your policies.
Overview
Many companies run reporting off a tangle of scripts, spreadsheet exports and a database replica. Numbers differ between teams, jobs fail silently, and nobody knows which tables hold personal data.
We build a lakehouse on S3, ADLS or GCS with Postgres or ClickHouse for serving, dbt for tested transformations and Airflow or Dagster for scheduling. Data is classified as it lands, and access is granted by role, not by sharing passwords.
Signs you need this
- Two dashboards show different figures for the same metric
- Pipeline failures are discovered when someone asks why a report is stale
- Analysts query the production database directly
- You cannot list which tables contain personal data or who can read them
What you get
- Platform architecture
- Storage layers, compute, serving databases and network design, written as infrastructure as code.
- Ingestion pipelines
- Batch and change-data-capture loads from your source systems into raw storage, with schema checks.
- dbt transformation project
- Modelled tables with tests, documentation and lineage, reviewed through pull requests.
- Orchestration
- Airflow or Dagster jobs with dependencies, retries and alerts to the people who own each dataset.
- Data classification and access control
- Tags for personal, financial and confidential data, applied to columns, with role-based access and masking.
How the work runs
- 1
Discovery
1 to 2 weeks
We list sources, consumers, key metrics and data sensitivity, and agree the first use case.
- 2
Foundation
2 to 4 weeks
We build storage, compute, networking, access roles and the first ingestion pipelines.
- 3
Modelling
3 to 5 weeks
We model the first domain in dbt with tests and connect it to a reporting tool.
- 4
Classification and handover
1 to 3 weeks
We tag sensitive data, apply masking and access policies, and train your team on the workflow.
Timings depend on the size of your estate. The assessment fixes them before any build starts.
What you can check afterwards
- Each reported metric is defined once in dbt and its lineage is visible
- dbt tests and pipeline alerts notify a named owner when data is late or wrong
- Columns holding personal data are tagged and masked for roles without clearance
- Analysts no longer have credentials to the production database
Tools we work with
- Amazon S3
- Azure Data Lake Storage
- Google Cloud Storage
- Apache Iceberg
- dbt
- Apache Airflow
- Dagster
- PostgreSQL
- ClickHouse
- Debezium
- Terraform
- Metabase
Questions buyers ask
- Does our data have to leave the UAE or the EU?
- No. The platform runs in your cloud account in the region you choose, and we use tools you host or that run in that region.
- Do we need a big data team to run this?
- No. One or two engineers who know SQL can run it day to day. We pick tools that fit your team size.
- Why not just buy a managed warehouse?
- Sometimes that is the right answer, and we will say so. Open table formats on your own storage avoid lock-in and can keep data in regions some vendors do not serve.
- What access do you need?
- Admin rights in a cloud account for the platform, read-only access to source systems, and time with the owners of key metrics.
Talk to us about data platforms and pipelines.
Tell us what you run and where. We reply within one business day with questions or a quote.