Hire a Databricks engineer

Whatever your team needs on Databricks: loading data from sources with Lakeflow Connect or Spark, building pipelines with Lakeflow, scheduling tasks in Lakeflow Jobs, managing permissions and lineage in Unity Catalog, or defining jobs and pipelines as code.

The engineer works as part of your team, on your backlog. You can read how selection and placement work on our staffing page. Our partner page explains how we work with Databricks.

A Databricks engineer knows the platform’s components: Delta Lake, where all tables on Databricks are stored by default, Unity Catalog, Lakeflow, serverless compute and SQL warehouses. That matters if you already use Databricks or are migrating to it.

If the work is mainly about making data sources accessible and building data models, regardless of platform, a data engineer with broader experience is a better fit.

Yes, provided the engineer has account admin permissions or works with your admin. Workspaces created after 8 November 2023 have Unity Catalog automatically. For older workspaces, you set up identities at account level, connect the workspace to a metastore, migrate tables, and update queries and jobs.

For larger migrations, Databricks Labs provides the UCX tools. Afterwards, you disable the old features so no one can bypass the new permissions.

By running each workload on the cheapest suitable compute. Scheduled pipelines run on job compute, which Databricks says costs considerably less than all-purpose compute. SQL workloads belong on a SQL warehouse.

The engineer also tags workspaces and clusters by team or project from the outset, because tags cannot be applied retroactively. You can then see usage by tag in the system.billing.usage system table.

Our Databricks engineers are employed by Data Science Lab, which places them with clients. You have one point of contact for selection, onboarding and follow-up, and the engineer can draw on colleagues with other specialisms.

With a freelancer, you assess the quality yourself. Since 1 January 2025, the Dutch Tax Administration has resumed enforcement against false self-employment. As the client, you must therefore assess whether the assignment is suitable for a self-employed professional.

An account in your identity system, such as Microsoft Entra ID, that is provisioned at account level in Databricks. Grant permissions through a group in Unity Catalog, limited to the catalogues and schemas needed for the work, so you can revoke them easily at the end.

Also arrange access to the repository and development environment, and document who the client is and which standards apply.

By defining jobs and pipelines as code rather than creating them manually in the workspace. With Declarative Automation Bundles, formerly Databricks Asset Bundles, jobs, pipelines and their settings sit alongside the code in your repository, with review and CI/CD.

Where it makes sense, the engineer works on the same tasks as an internal colleague. Knowledge transfer happens when you want it, not just at the end of the assignment.

If the work is ongoing, you may be better off hiring someone directly. If you want a completed platform rather than additional capacity, a project is a better fit. If the issue is how data is defined or who owns it, more engineering will do little; start with data governance instead.

Want to compare providers first? See our overview of Databricks partners in the Netherlands.