Asking company data questions in plain language

The problem

Data at a large automotive software organisation sits in many separate sources, and non-specialists struggle to get answers from it. My master’s thesis (University of Ulm, in cooperation with Volkswagen / CARIAD) addressed this with a centralized platform and an AI interface on top.

My part

• Centralized data platform on Databricks.
• ETL pipelines ingesting data from various sources, so that key information is available in near real time.
• Machine learning models for platform functionality, including text-to-SQL queries and natural-language data exploration.
• Requirements defined together with stakeholders to stay aligned with business objectives.

Result

A working platform where users can explore data in natural language instead of writing SQL, delivered as the basis of my MSc thesis (April 2025).