Join the team
We're a small, research-led team building independent security testing for AI agents. We're hiring engineers and interns, and running the Sentinel Fellowship for researchers. All roles are hybrid.
Who we're hiring
Click a role to see what it involves.
Full-time Software Engineer
Build the platform that tests and protects our customers' AI agents.
What you'll do
- Build the systems that run large-scale security tests against AI agents and report what they find.
- Build the runtime gateway that checks an agent's high-risk actions before they execute.
- Integrate with the tools, APIs, and agent frameworks our customers use.
What we look for
- Strong software engineering experience, especially backend systems and APIs.
- Care for reliability and security in the code you ship.
- Interest in AI and LLMs. Prior ML experience is a plus, not a requirement.
Full-time AI Engineer
Turn AI security research into automated tests that run on real agents.
What you'll do
- Develop automated red-teaming methods for LLMs and agents, such as jailbreaks, prompt injection, and tool misuse.
- Build evaluation pipelines that measure how safe an agent is, and how that changes over time.
- Work with the research team to bring new attacks and defenses into the product.
What we look for
- Hands-on experience with LLMs: training, fine-tuning, or evaluating them.
- Strong Python and experience with PyTorch or similar frameworks.
- Experience in AI security or safety research is a plus, as are publications.
Internship Engineering and Research Intern
Spend 3–6 months on one focused project, in engineering or research, with a founder as your mentor.
What you'll do
- Own a well-scoped project, such as a new attack technique, an evaluation benchmark, or a platform feature.
- Work alongside the team on real systems, not a side project.
What we look for
- Students in computer science, machine learning, security, or a related field.
- Solid programming skills and curiosity about how AI systems fail.
A research fellowship in AI agent security
The Sentinel Fellowship supports researchers working on open problems in AI agent security. Fellows spend six months on one research project, mentored by our founders, with the goal of a public result: a paper, an open-source tool, or a benchmark.
Research areas
Red-teaming AI agents, runtime guardrails, neuron-level safety and repair, and how to tell when a system has been tested enough to deploy.
What fellows get
A £1,000 stipend for the six months, close mentorship from researchers who publish at top security and ML venues, and support to publish the result. Hybrid.
Who it's for
PhD students, postdocs, and engineers with a research background who want to work on AI security problems that matter in practice.
Work on problems that don't have answers yet
Research-led
Our founders publish at USENIX Security, NDSS, NeurIPS, and ICML. You'll work directly with them, and new research goes straight into the product.
Real stakes
AI agents are being given access to money, data, and internal systems. The question we work on is whether they can be trusted with it.
Early and small
You'll own whole parts of the system, not a ticket queue, and help shape how the company and product grow.