Founding AI Engineer

ManticoreAI. Backed by leading US cybersecurity investors. Remote. High-intensity.

We're an AI-native offensive security company, backed by well-known US cybersecurity investors, building what the security industry has talked about for a decade and never shipped. Pentesting today is slow, manual, and stale the day it's delivered. We rebuilt it from the ground up: AI agents that plan, exploit, pivot, and report against real customer systems the way an elite operator would, and prove what's actually exploitable instead of pattern-matching.

This is the engine. Not a CVE scanner, not a vuln-feed aggregator. A multi-agent system that runs a real penetration test end to end, then produces CREST-certified, audit-grade findings that Big 4 auditors accept for SOC 2, PCI DSS, and ISO 27001. We have a five-year head start on the architecture behind it.

20 months of build. Production platform. Paying customers. We're hiring our Founding AI Engineer to own, harden, and scale that engine alongside the founder. The platform is already live and working. Your job is to make it more capable, more reliable, and faster, not to start from a blank page. This is a real shot at defining a category.

Read this first. This is not a balanced-lifestyle role. We move at startup pace because the window to win this market is now, not in three years. If you want predictable hours, weekends fully off, and a calm Slack, this isn't it. If you want to take production AI agents that already break into real systems and push them further, alongside people who care obsessively about what they ship, keep reading.

What You'd Own

The engine already runs against live customer targets. You'd take ownership of it end to end and keep making it better, working directly with the founder:

• Multi-agent AI that reasons through an attack the way a senior pentester would, plans its next move, and adapts as it goes
• Real exploitation against live targets, not signature matching, including web, API, and network paths
• Browser-driven testing for modern web apps
• Post-exploitation and lateral movement once a foothold is established
• Reliability, cost, and speed of the whole agentic pipeline at production scale

This is not a write LLM prompts role, and it's not build-from-scratch. It's a take production AI agents that already exploit real systems and make them sharper role.

Who You Are

Must have:

• 6+ years shipping Python async in production (real asyncio, not just typing async def)
• Real offensive security background — OSCP / OSWE / OSEP or equivalent war stories. You can read a pentest report and know where the real finding is.
• Built or heavily modified LLM agent systems in production
• Comfortable owning and navigating large, complex codebases
• Strong systems debugging: async deadlocks, event-loop issues, process trees, signal handling

Nice to have:

• Hands-on with offensive tooling (the standard Kali toolkit)
• Prompt engineering at scale (many interdependent prompts in production)
• Postgres experience
• AI safety or red-teaming AI systems
• Previous startup or founding-engineer experience

Filters:

• You use AI tooling daily (Claude Code, Cursor, or equivalent). If you think AI tools make engineers worse, this isn't for you. We live in this stack.
• Strong written English (async-first team across timezones)
• You read documentation and update it. Our internal docs are deep, and how you use them in the take-home tells us a lot.

Tech Stack

Python and asyncio at the core, modern LLM tooling, and the standard offensive-security toolkit on Kali Linux. You'll work against a shared platform data layer. We'll go deeper on the architecture once we're in conversation and under NDA.

What This Isn't

• A ticket-taking role. You decide what ships next in the engine. You argue with the founder.
• A research role. We ship. Agent papers are inspiration, not the deliverable.
• A prompt engineering role. Prompts are part of it. The harder work is tool integration, agent orchestration, cost management, and reliability.
• A balanced-lifestyle role. We work hard. If that drains you, this isn't a fit. If it energizes you, you'll thrive here.