JOB-1715
A seed-stage startup that stress-tests, certifies and insures AI agents for enterprises is looking for senior engineers. The goal is to make the risks of agents deployed in sensitive settings measurable, so that businesses can adopt them with confidence. Engineers here lead adversarial evaluation of widely used agents, engaging hands-on with the agents customers have built, from the first scoping conversation to final certification.
Mission of the role
- Improve the central engine behind all of the company’s adversarial testing and evaluations
- Own each engagement as the directly responsible individual, from scoping through recon, testing, fixes and certification
- Turn cutting-edge attack techniques into repeatable product features
- Review evaluation output with a critical eye and decide what counts as good enough
- Automate your own workflow with agents
- Work out how to evaluate new kinds of agents, such as those that operate computers
Strong fits look like
- Five or more years as a software engineer owning high-stakes technical work end to end (exceptional earlier-career candidates may be considered)
- Strong Python and TypeScript, and experience shipping user-facing features to production
- A CS or similar technical degree and a rising trajectory in title and scope
- Daily, native use of agentic AI tools, with comfort around agent frameworks and evaluations
- Clear communication with both technical and non-technical audiences
- Founder, founding-engineer or early-startup experience is strongly preferred; this is an individual contributor role, not a management one
On-site in the San Francisco Bay Area, five days a week. Visa transfers considered.
