Sr. Principal Scientist, Kiro Science

Description
We are building the systems foundations for the next generation of autonomous coding agents, agents that operate in parallel, run for hours or days, and collaborate on codebases at a scale that breaks the assumptions of tools built for human developers. The hard problems in this space are no longer about model quality alone. They are about the systems layer underneath: how agents coordinate, how state is managed across long horizons, how correctness is preserved when many agents act concurrently, and how deterministic infrastructure can take on responsibilities that LLMs cannot reliably handle.

Our team treats these as first-class research and engineering problems. We are designing new primitives for coordination, concurrency control, state management, validation, and inter-agent communication that move responsibility out of fragile LLM-mediated paths and into systems that are predictable, debuggable, and fast. Early prototypes have produced order-of-magnitude i...