Search is becoming the core primitive for AI. Every agent workflow, every reasoning system, every tool call needs context from the web and beyond, but today's search infrastructure was built for a different era. We're building the stack that makes the web accessible to how agents actually work: content at scale, structured for reasoning, designed for deep research across thousands of sources. Information Retrieval is in a renaissance, LLMs are changing not just what we need to build but how we can build it, and we combine first-principles research with hard engineering to solve web-scale indexing for agentic use.

How agents access the web is one of the foundational questions of the next decade of software, and the infrastructure is what makes the answer fast, cheap, and possible at scale. You'd be building the layer underneath everything: storage, retrieval, indexing, serving. The kind of work that ends up in papers and the kind of work that ends up running in produc...