About the role
The offensive security team attacks our own systems before anyone else does. We run red team engagements against cloud estates, internal networks, endpoints and the sites that house the lidar test fleet. We test the AI systems too: model weights, inference servers, evaluation pipelines and agent permissions. Every engagement ends with a reproducible attack path and a fix that the owning team has agreed to.
Frontier AI and Autonomy & Sensing hold the parts of Clexerl that are hardest to replace: trained weights, safety cases and sensor data. Adversarial work here covers prompt-based abuse of agents, extraction of weights through inference APIs, and spoofing of perception inputs such as lidar returns. Results feed directly into the safety cases that gate releases, and into the permission model of Clexerl Go, where agents act on people's behalf.
Engagements are scoped in writing with agreed rules of engagement, and we tell the owning team when we start. We value a clean report that a busy engineer can act on within a day.
What you’ll do
- Plan and run red team engagements against cloud accounts, internal networks and applications, within agreed rules of engagement.
- Develop proof-of-concept exploits and attack chains that show real impact, then hand them to the owning team with a fix.
- Test model-serving infrastructure and agent tool permissions for escalation paths, including prompt-driven misuse.
- Simulate adversarial inputs against perception pipelines, such as spoofed lidar returns and manipulated camera frames.
- Write findings that a non-specialist can act on, with severity, reproduction steps and a recommended fix.
- Build reusable attack tooling and detection test cases, and share them with the security detection team.
- Track remediation through to verification, retesting each fix before a finding is closed.
What you’ll bring
- Experience in offensive security, penetration testing or red teaming, with real engagements delivered and reported.
- Deep knowledge of Active Directory, cloud IAM models and common privilege escalation paths on Linux and Windows.
- Strong web and API testing skills, including authentication bypass, server-side request forgery and business logic flaws.
- Ability to write exploit code or tooling in Python, Go or C, and to read unfamiliar code quickly and accurately.
- Clear written reporting, with reports that engineers use rather than file away.
- Judgement about scope and safety, including knowing when to stop and escalate during a live engagement.
Nice to have
- Experience attacking machine learning systems, including model extraction, membership inference or data poisoning.
- Hardware or embedded security experience, such as firmware extraction, fault injection or CAN bus analysis.
- Industry certifications such as OSCP, OSEP or CRTO, or equivalent evidence of practical skill.
- Experience presenting findings to executive audiences without losing the technical substance.
Work you might take on
- 01
Run a red team exercise against an agent deployment, testing whether crafted tool outputs can escalate its permissions.
- 02
Build a harness that tests spoofed lidar returns against the perception stack and records how the planner responds.
- 03
Write an attack-path analysis for the cloud estate that maps identity trust relationships and ranks the paths by impact.
- 04
Produce a reusable test suite for model-serving endpoints covering authentication, rate limits and extraction attempts.
Compensation
$250K – $400K
The US base salary range for this role. Every offer also includes equity in Clexerl and, when it spins out, in the venture you help build. We tell you the reasoning behind your number.
Benefits
Health, fully covered
Medical, dental and vision for you and your dependants, premiums paid in full.
Two kinds of equity
Equity in Clexerl, plus a stake in the venture you help build when it spins out.
22 weeks parental leave
Fully paid, for every parent, however your family arrives.
Rest that sticks
Flexible time off with a four-week minimum we actually check on, plus a company-wide week in December.
Learning and compute
A yearly budget for courses, conferences and books, and compute credits for side projects.
401(k) with a 4% match
Matched from your first paycheque and vested immediately.
How we hire
Apply 5 minutes
Send links to what you have built and a few lines on why this role. No cover letter needed.
Intro call 30 minutes
A conversation with the hiring manager about the work, the team and what you want next.
Work sample 2–4 hours
A paid take-home or a pairing session on a real-shaped problem. You pick the format.
Studio day One day
Meet the team, go deep on your work and ours. Remote or in person.
Offer Within 5 days
A written offer with the full range, the equity and the reasoning behind both.
Opens an email with a short template. Links beat résumés.