DeviantArt, a Wix subsidiary, is one of the world's largest online art communities, with more than 100 million registered members and over 650 million works spanning digital art, photography, illustration, animation, sculpture, and emerging media. Helping users discover relevant content across this vast and continuously evolving catalog is fundamental to the platform's user experience.
Every day, tens of thousands of new artworks are uploaded by creators around the world, creating a constantly changing mix of content, trends, and user interests. Supporting discovery at this scale requires far more than traditional search. DeviantArt must retrieve and rank content in real time, understand artistic styles and community terminology, personalize results for diverse audiences, and deliver responsive search and recommendations despite continuously changing traffic patterns.
At DeviantArt, discovery is fundamental to the user experience. With hundreds of millions of artworks, rapidly changing content, and highly diverse user interests, we needed a platform capable of combining machine learning, personalization, and real-time serving performance at a very large scale. Vespa gives us the flexibility to integrate custom ranking models, user context, and domain-specific processing within a single high-performance system, which is critical for delivering responsive and relevant experiences across our global community.
DeviantArt uses the Vespa AI Search Platform to combine machine learning, custom query processing, document enrichment, and real-time ranking within a single serving layer. Vespa’s architecture enables the platform to support advanced personalization, custom linguistic processing, and flexible ranking models while maintaining the low latency and operational scalability required by one of the world’s largest creative communities.
Today, the deployment manages more than 1 billion documents and serves nearly 4,000 queries per second, with an average query latency of 12 milliseconds. At the same time, the platform continuously ingests new and updated content at sub-millisecond feed latency, enabling DeviantArt to deliver fast, highly personalized discovery experiences across one of the world’s largest and most active online art communities.
By using Vespa.ai as an integrated AI-powered search and discovery platform, DeviantArt can combine machine learning, personalization, custom query processing, and real-time ranking within a single high-performance serving layer, supported by key capabilities that include:
Custom query processing and user context support individualized search and recommendations.
Custom linguistics and document enrichment help Vespa understand artistic terminology, tags, styles, and metadata.
Custom ranking models and ONNX inference combine semantic, behavioral, and structured signals at serving time.
Autoscaling, Enclave isolation, parent/child models, and custom components support a secure, adaptable platform at global scale.
Whether you're building AI search, recommendations, RAG, or AI agents, we'd be happy to discuss how the Vespa AI Search Platform can help you deliver fast, personalized AI experiences at production scale.
