Architect Of Storage Solutions (all Genders) - 6 Weeks Freelance Contract at Wikimedia Deutschland E. V.

• Job Location Berlin, Berlin, Germany
• Seniority Mid-Senior Level
• Type Contract

Description

Wikidata is Wikimedia’s knowledge graph, which acts as a backbone for Wikipedia and other Wikimedia projects.

It is also a significant part of the Linked Open Data network, and through being publicly available allows anyone to access its data.

Wikimedia Germany is looking for a storage solution consultant to help us evolve how Wikidata stores and provides its data in the decades to come.

Wikidata is a wiki that everyone can edit, either manually on wiki pages or programmatically.

Data from Wikidata’s graph can be accessed in a number of ways: through various web APIs, dedicated SPARQL API and querying UI, or data snapshots (data dumps) provided periodically.

Project details Seniority: senior level Starting Date: Mid February 2026 Duration: 4-6 weeks / we're not planning to hire on a permanent basis Hours per week: 30-40 hours / week Location: Germany / remote Scope We are seeking an experienced architect of storage solutions on a freelance contract to liaise with the product development, SRE and platform teams that support the existing system in order to analyze and ideate potential approaches for data storage of Wikidata to support its strategic goals and growth in the period 2026-2035.

Technical background information Primary data storage used by Wikidata is a Mediawiki relational database (Maria DB) storing data objects in the form of string representations of JSON objects.

Several secondary storage approaches have been introduced optimized for particular use cases Dedicated SQL table dedicated to labels (“titles”) Dedicated SQL tables storing “links” between different elements of the knowledge graph Elastic Search index for search RDF Triplestore that enables SPARQL querying Wikidata primary data storage in numbers (state Jan 2026) Database size: 1.2 TB (900 GB in Jan 2025) Average rows read rate: 1.98 M read rows/second Average rows written rate: 5K written rows/second Examples of known limitations and risks of the current storage approach The SQL tables storing data of all versions, and their relevant metadata, of Wikidata data objects has been growing too big to be efficiently served: For every version of a data object there is an entry stored permanently in the respective table.