Activeloop was founded in 2018 by Davit Buniatyan, Sergiy Popovych, and Jason Ge, going through Y Combinator the same year and later raising an $11 million Series A backed in part by Streamlined Ventures. The company is based in Mountain View, California, and built Deep Lake, an open-core database designed specifically for the multimodal, unstructured data that deep learning and LLM applications consume.
Deep Lake stores embeddings, images, audio, video, and text in a format built for streaming directly into model training and retrieval pipelines, with built-in dataset versioning and visualization so teams can track how their training data changes over time rather than treating it as a black box. The open-source core (7,000-plus GitHub stars) means teams can self-host, while the managed service adds hosted storage, API access, and enterprise support on top.
Activeloop targets ML engineering teams building retrieval-augmented generation, computer vision, or agent systems who need dataset infrastructure beyond what a standard vector database or object store handles, with disclosed customers including Intel, Google, Matterport, and Bayer in regulated industries like biopharma and automotive. The free tier caps out at 100MB storage and 3 queries/day; the Pro plan runs $40/month per user with metered overages, and Enterprise pricing is custom.
