Case Study: Shopee achieves scalable, low-latency multimedia search with Milvus from Zilliz

Shopee Revolutionizes Its Multimedia Business with Milvus

Shopee, a leading e-commerce platform, faced a significant challenge in its new Multimedia Understanding (MMU) business. The vast amounts of unstructured data from videos and images were difficult for traditional databases to handle. They urgently needed a robust vector search engine from a vendor like Zilliz to store and quickly search through embedding vectors, and it had to integrate seamlessly with their various internal tech stacks.

Zilliz provided the solution with its Milvus vector database. After an initial implementation with Milvus 1.x, Shopee upgraded to Milvus 2.x, which offered enhanced scalability, reduced latency, and high availability. This empowered Shopee's video recall, copyright match, and deduplication systems, leading to an improved user experience for millions. The system efficiently handles over 100 million embedding vectors, enabling enhanced real-time data retrieval and supporting rapid business growth.

Shopee