A serverless vector database built for AI at scale.

Vectroid is a serverless vector database that lets AI teams store billions of vectors and query them with extremely low latency, with fast indexing, flexible APIs, and enterprise-grade encryption built in.

Vectroid was founded in 2024 by Talip Öztürk, Kevin Hanson, and Zoltan Baranyi. Talip previously founded Hazelcast, the open-source in-memory computing platform, later founded the cryptocurrency wallet and exchange Evercoin, then moved into engineering leadership roles at Google, scaling the distributed relational database AlloyDB, and at Vectara, a GenAI retrieval-augmented generation platform. Zoltan, who built Vectroid's storage engine and search algorithms, spent over six years at Hazelcast before that, engineering its tiered storage and WAN replication systems. Kevin led Vectroid's go-to-market as COO, drawing on a database and cloud sales background spanning MongoDB, Google Cloud, and Grafana Labs.

As AI applications moved from prototypes into production, the vector database layer became a real bottleneck. Vectroid's founders combined infrastructure engineering depth with a product built specifically for the scale AI workloads demand, and Race Capital's early conviction (an early Databricks investor) reinforced our own.