Human Archive
Multimodal data provider for robotics and world modeling
What they do
Robotics and world-modeling teams lack massive, diverse, high-quality real-world multimodal datasets needed to train robust embodied intelligence systems.
The company collects aligned multimodal data using custom hardware in real environments and delivers it after QA, anonymization, and annotation as scalable datasets for training robotics foundation models.
Frontier AI labs and general-purpose robotics companies developing robotics foundation models and embodied intelligence.
Founders
Archiving the structure of human interaction in the physical world. Berkeley dropout and previous farmer (sold mangoes & planted trees)
Rushil Agarwal is a founder of Human Archive (W26). He studied Industrial Engineering and Operations Research along with Business at UC Berkeley's highly selective MET program (<1% acceptance rate) He previously worked on data pipelines for embedding models and on-chain infrastructure at Coinbase. He also scaled a Web 3.0 software business, which served 500+ SMBs
Creating multimodal real-world datasets for robotics
Shloke Patel is a founder of Human Archive (W26). He studied Mechanical Engineering and Computer Science at Stanford University, where he focused on robotics, systems engineering, and applied AI, and conducted research at the SHAPE and DEL Labs. Before Human Archive, Shloke built and operated agricultural businesses, including a mango venture that sold 16,000+ mangoes and helped plant 100,000+ trees. His background spans robotics, mechanical design, and hardware systems.
Open roles 8
Company timeline
- Founded2026
- YC batchWinter 2026
- StatusActive
- Team size4 people
- LocationSan Francisco, CA, USA
Similar startups
