AI Offensive Security Engineer
Run pentests and red team engagements against apps, APIs, cloud and AI features, and turn our offensive methodology into a standard.
About Secure Channel
Attackers adopted AI faster than the companies they target. Secure Channel helps them catch up. We find where AI has left a company exposed, tell them what to fix first, and help them fix it.
We focus on the security problems AI created. We don't take on the old problems that got 50% harder. We take on the ones that got 10x or 100x harder, or didn't exist before: prompt injection, agents with tool access to internal systems, MCP servers wired into production, and products sending customer data to models.
We sell fixed-price, time-boxed work: two-week sprints and scoped penetration tests. Every finding comes with proof and a fix. Our founders have spent their careers building and breaking systems at Bishop Fox, Snap, DARPA, Oak Ridge National Laboratory and a string of startups.
The role
You'll run penetration tests and red team engagements against our clients' applications, APIs, cloud environments and AI features. AI will do real work in how you test.
We already have methodologies, tools and agent skills that work. Your job is to make them a standard: a repeatable, AI-assisted offensive process that delivers the same quality every time, no matter who at Secure Channel runs it.
This role is meant to grow into leading our offensive practice.
What you'll do
- Run pentests and AI pentests end to end: scoping, testing, exploitation, reporting and the client readout.
- Attack AI systems the way a real adversary would. That means prompt injection, agent and tool misuse, MCP abuse, and reaching data or actions a feature should never allow.
- Turn our offensive methodology into a standard: tooling, agent skills, playbooks and report templates.
- Write findings with proof: reproduction steps, real impact and a fix the client can ship.
- Own client work, including public sector engagements, without a founder in the room.
- Find new attack classes and exploits worth publishing.
What we're looking for
- 5 to 7 years in security, with real offensive work: pentesting, red teaming, exploit development.
- You're a builder. You've worked as a software engineer in at least one role and shipped production systems. You know how software gets built inside a company because you've done it. Scripts that help you test don't count.
- You know how AI systems are built, so you can break them. That includes LLM application architecture, agents and tool calling, MCP and related protocols, and prompt injection.
- You work AI-native. You use coding agents and AI tooling every day, and you've built your own.
- You take security seriously, and it shows. CVEs, bug bounties, CTF results, offensive certs (OSCP, OSWE, OSEP, OSCE3), conference talks, open-source tools. Any of these counts.
- U.S. citizenship. Some of our work is in the public sector. No clearance required.
Nice to have
- A consulting background, meaning you've written reports and run client readouts.
- Early-stage startup experience, ideally both building products and breaking them in production.
This isn't the role for you if
- Your background is mostly blue team, SOC or security analyst work.
- Your credentials are mostly network or vendor certifications.
- You'd rather analyze and recommend than build and break. We hire founder types: people who own the outcome and do the work themselves.
Logistics
- Location: Remote, U.S. We're based in Venice, CA. Bonus points if you're near Los Angeles, East Tennessee (Knoxville, Oak Ridge, Tri-Cities) or Washington, D.C.
- Travel: Minimal, roughly 1 to 2 weeks every six months.
- How we hire: You start as a contractor on one or two client engagements, paid per engagement (hourly or a fixed fee per sprint). If it's a fit, we convert you to full-time W-2.
- Compensation: $100K to $180K base salary once you convert to W-2, plus health insurance.
How to apply
Email chris@securechannel.ai with:
- A short introduction
- Your resume
- Links to your work: CVEs, write-ups, talks, repos, tools, products you've shipped