Hungarian commercial bank
Data & AI Engineering
Data warehouse migration to a Databricks lakehouse
Our role: 4-person data engineering team · technical delivery of the migration with the bank's data team
DetailsArea 01 — Data & AI Engineering
We provide reliable data for reports, applications and AI.
AI and reporting are only as good as the data underneath. We build and modernise data platforms, data models and pipelines on Databricks, Snowflake and BigQuery – so that your in-house team can run them confidently afterwards.
Area 01
Sound familiar?
Head of Data / Data Platform
Platform and operations
- Most of the platform team's time goes on fixing pipeline failures and ad hoc data requests.
- Data issues are usually spotted by the business, not by monitoring.
- Platform cloud costs keep rising, but nobody can see which domain or workload drives them.
- The legacy warehouse and the new lakehouse have run side by side for years; the migration never closes.
CDO / Data Governance
Data quality and compliance
- Different reports give different numbers for the same KPI.
- Lineage for regulatory reports has to be partly rebuilt for every audit.
- Governance rules are documented but not technically enforced in the pipelines.
- Nobody can quantify the business value of the data platform budget.
Head of AI / CTO
AI-ready data
- AI use cases stall after the PoC on data availability and quality.
- Contracts, documents and correspondence sit unstructured, without consistent access control.
- AI teams build their own data copies, creating uncontrolled parallel pipelines.
- There is no shared data model or catalogue that several AI applications could safely build on.
How it works
A typical enterprise data flow from source systems to reports and AI.
Data is ingested from source systems, stored raw, then transformed and validated. In modelled form it reaches reports, applications and AI. A scheduler runs the whole flow, its operation is monitored continuously, and every step has an owner in the team.
DEData Engineeringestion, pipelines, transformation
AEAnalytics Engineerdata tests, data model, business logic
PEPlatform Engineerscheduling, infrastructure, monitoring
AIAI EngineerAI applications, RAG
Data & AI Engineering
How we help
01
Data platform build
We design and build lakehouse and data warehouse architectures on Databricks, Snowflake or BigQuery, from access control to development environments.
02
Ingestion and integration
We connect source systems (ERP, CRM, databases, APIs, files) with batch, CDC-based or real-time ingestion.
03
Pipelines and ETL/ELT development
We write tested, version-controlled transformation pipelines in Spark, SQL or dbt that turn raw data into reliable, business-ready tables.
04
Scheduling and automation
Pipelines run in dependency order with Airflow or Databricks Workflows. Failures trigger alerts; reruns and deployment (CI/CD) are automated.
05
Data modelling
We organise data into layered, dimensional models with shared business definitions and metrics that both reports and AI applications can use.
06
Migration
We replace legacy data warehouses and ETL tools (Informatica, SSIS, ODI) with parallel runs and reconciled results.
07
Data quality and lineage
Automated quality checks, lineage and named data owners: it is clear where a number comes from and who is responsible for it.
08
Performance and cost
We tune slow jobs and queries, remove unneeded compute and make platform costs measurable.
Data & AI Engineering
How we start
Assessment
Data platform and AI feasibility assessment
For one selected business need, we look at which data and integrations exist, what is missing, and what a realistic path to delivery looks like.
- What we start with
- One specific business need and access to the relevant systems
- What you get
- A data and integration map, a list of risks and a step-by-step plan
- Who works on it
- An experienced data engineer and architect, with your data team
Scoped project
Fixing one critical pipeline
We make a single business-critical pipeline or data product reliable: data model, quality checks, monitoring, documentation.
- What we start with
- A faulty or slow pipeline that feeds a report or application
- What you get
- A stable pipeline with measured quality, handed over to your in-house team
- Who works on it
- 1–3 experienced engineers, with handover criteria agreed in advance
Technology
Data platforms, data modelling, pipelines, migration, data quality, performance and cost optimisation.
Platforms
Databricks, Snowflake, BigQuery
Processing
Apache Spark, SQL, Python, dbt
Orchestration
Airflow, Databricks Workflows
Operations
Docker, Kubernetes, CI/CD
Case studies
All case studies Hungarian retail chain
Data & AI Engineering
Unified data model and Snowflake cost optimisation
Our role: 2 data engineers · alongside the internal BI team
DetailsFAQ
Data & AI Engineering
Do you only work with Databricks, Snowflake or BigQuery?
That is where most of our experience is, but we always start from your existing systems. We only recommend a new platform if there is a business reason for it.
Do we have to replace our existing data warehouse?
Not necessarily. It is often faster to fix the existing platform and replace only the bottlenecks.
Who runs it after the project?
By default, your team. Handover includes documentation, monitoring and a period of joint operation.
Can you work alongside our in-house data team?
Yes, that is how we usually work: with your tools, in your codebase, to your processes.
The first step
Work with us
Bring a specific data or AI problem.
In 30 minutes, one of our experienced engineers talks through the situation with you and what it would take to solve it. Not a sales pitch.
Data platform and AI feasibility assessment
For one selected business need, we look at which data and integrations are needed, what is missing, and in what steps it makes sense to deliver it.
Enterprise knowledge search or impact analysis
We deliver one well-defined use case, with data sources, access rights and quality expectations agreed in advance.