We're a small, fully remote team. AI agents are already running in production codebases, making commits, calling APIs, and touching production systems. We're building the security layer that goes underneath them.
AI coding agents open pull requests, call production APIs, and install packages before anyone approves a step. They inherit every credential the developer holds, and one malicious instruction buried in a file they read can turn them against you. Most teams ship them anyway with the safety flags off, because the alternative is sitting beside the agent all day clicking approve.
Our job is to give every agent a security layer it can't switch off and no one has to babysit. Policy enforcement, secret protection, and prompt-injection detection, wired in before the agent runs its first command. When it works, the developer forgets it's there. The agent moves fast and the security stays on.
Security is the product here. Every call we make starts from one question: would a security engineer with a good gut let this through? That keeps us close to the edge of what agents can do, which means a lot of reading, a lot of breaking things on purpose, and sharing what we learn in the open.
We're fully remote. You own your work and your time. Monthly bonuses for meeting fitness goals!
We run lean on purpose. A small team that communicates well and uses AI agents to the fullest can outship teams ten times larger. We add clarity, not headcount.
We build the tools that secure AI agents, and we use those same agents to build faster. Code review, research, security scanning, documentation. Staying at the edge of what agents can do is part of the job.
We don't always have open roles posted. But we're always interested in exceptional people. Tell us about yourself and what you'd want to work on. We read every submission.