Snatch Up

Vulnerability Researcher | New York (in office, 5 days/week) | Full relocation support

A small, well-funded startup is building something new: systems that use AI to find vulnerabilities and develop exploits at scale. The team moves fast and works close to the metal. It's made up of people who've done serious work before, including alumni of frontier AI labs, top US intelligence agencies, elite Israeli intelligence units, and a defense tech unicorn, backed by tier-1 US investors..

This is not a prompt engineering job. You'll work at the level of OS internals and kernel interfaces, optimize for real hardware, and ship production code where performance and correctness genuinely matter. You'll own problems from first idea to running system.

What you'll do

• Build production systems that drive AI-powered vulnerability research and exploit development
• Work in OS internals, kernel-level interfaces, and low-level performance optimization
• Own problems end to end: architecture, implementation, and shipped code
• Partner with vulnerability researchers and security operators to turn slow manual cyber work into automated, repeatable systems
• Help set the standards and rhythm of a young team aiming to lead AI-native offensive cyber

What you'll need

• ~2+ years of engineering experience (exceptional new grads with real low-level work are encouraged to apply)
• Deep low-level systems experience: OS internals, kernel work, runtime detection, dynamic analysis, dynamic interception, or similar
• A builder mindset: comfortable owning systems end to end and shipping fast, correct code
• Real comfort with ambiguity. This is the most important thing
• Willing to be in person in Washington, D.C. five days a week (full relocation support)

Nice to have

• Vulnerability research, CTF, or offensive security experience
• High ranking on ctftime.org or similar
• Published research or notable CVEs
• Founder or early-engineer experience
• Open source contributions to OS-level tooling
• Experience at a cyber/security startup (runtime detection, dynamic analysis, OS tooling)
• Exposure to ML or AI systems work



If jumping on a flight across the world on short notice sounds exciting rather than stressful, you'll fit right in. This is for the engineer who doesn't just want to join a great company, but help build one.