Our data engineering consulting is for companies that want their data on a platform that grows with the business. We build the infrastructure, pipelines and integrations that analytics and AI workloads can run reliably on. We have built platforms from scratch and moved teams off Y42, a home-grown warehouse or a legacy ERP onto a modern data stack.
“In a migration I don't just rebuild every connector one to one. I first look at what the old logic actually does. In the end your team should be able to change the pipelines itself.”
Lui
Lead Data & AI EngineeringEvery data product is only as good as the platform under it. Four building blocks make up a Modern Data Stack that holds up.
We build your cloud data warehouse, for example in Snowflake or Google BigQuery. With dbt we model, test and document the data, so one single source of truth emerges for reporting and analysis. You get a warehouse with dbt models in staging, intermediate and mart layers, with tests and documentation.
In data integration, data from your shop, marketing, CRM and support tools flows into the warehouse automatically and on schedule. Standard sources such as your shop and ad platforms come in through Fivetran. For sources without a ready-made connector we write our own pipelines with dlt and Airflow, including SOAP interfaces or an ERP database reached over VPN. Failed runs report to a dedicated alerting channel.
When standard tools do not fit, we build custom data apps and internal tools (sometimes called a Company OS) that your whole organisation keeps coming back to. Sometimes the warehouse needs to produce a finished output. For Tasty Urban the pipeline generates the credit notes for partner restaurants and creates them as drafts in the invoicing tool. For another client, Claude queries the data directly through a semantic layer.
We set up the cloud infrastructure under your data platform. Servers, network and access are described as code with OpenTofu or Terraform and Ansible, so the setup is versioned like code and can be rebuilt on demand. Secrets live in a password manager, never in the repository.
Code and accounts belong to you from the start.
Three clients, three data platforms: this is what data engineering consulting looks like at Gemma Analytics. Two of these projects are DWH modernisations, the move off Y42 and the switch from a historically grown tool landscape to a modern data stack.
Fivetran and Airflow load every source into BigQuery, and dbt models the data. The platform was rebuilt step by step.
Pets Deli moved from a grown tool landscape to a modern, dbt-tested data stack on Snowflake.
Order and cost data from every source flow into a warehouse and credit-note app. Credit notes are drafted automatically, Finance reviews, EasyBill pays out.
Does any of this sound familiar? Then your data platform is ready for a closer look.
Shop, marketing, CRM and support each deliver their own numbers, and fragmented, manual processes make it hard to track key metrics.
The setup stopped scaling as the business and its data volumes grew. Every new question means another tab. By now a data warehouse modernisation is overdue.
When refreshes fail, trust in the numbers suffers, and the team goes back to checking everything by hand.
Loading, transformation and orchestration all depend on it.
Your reporting has to keep working, so the data migration to the new data platform has to happen without a gap.
One person built it in-house over years, and only they understand it.
We work hands-on and results-driven, in four steps.
We spend 30 to 60 minutes on your challenges, your current setup and where your data lives today.
We clarify which sources and systems to connect, design the target architecture for your warehouse and write down for each building block when it is done.
We build pipelines, warehouse and data models step by step: loaded automatically, tested and documented. Every week we show you the current state, and every few weeks we look at budget and next steps together.
The platform runs in production. Handover includes documentation and recorded hands-on sessions in which your team changes a connector itself and opens a pull request. Where it helps, we also cover interim roles in your data team.
We work with these technologies, among others, from Snowflake consulting to dbt and Airflow. Which ones fit depends on your setup.




70+ successful projects in six years, a senior team of 20 based in Berlin.
Book a free discovery call: 30 to 60 minutes, focused on your challenges.