Security Researcher

Collision AI is building models and agents for cyber defense. We’re looking for a security researcher to explore what these systems can do on difficult, real-world security problems. You’ll build agents, use existing models and tools in your own research, and turn the approaches that work into capabilities that help defend our customers.

You’ll develop new methods for agent-driven vulnerability discovery, exploring complex codebases and validating whether findings are exploitable. You’ll combine deep security knowledge with hands-on engineering: connecting agents to security tools, giving them the context to investigate systems, and checking their findings against reproducible evidence. You’ll study where they succeed and fail, build evaluations around realistic tasks, and work with our model researchers to improve their capabilities.

As an early member of the team, you’ll help shape our security research agenda, the systems we build, and the standards we use to judge whether an agent can be trusted with real security work.

• A strong background in security research, with depth in an area such as vulnerability discovery, application security, reverse engineering, or detection and response.
• Strong programming skills and the ability to build tools, automate investigations, and work with unfamiliar codebases and systems.
• The ability to investigate open-ended security problems, test hypotheses, and produce reproducible findings.
• Experience building LLM-powered agents or using models to carry out substantive security research.
• Experience with fuzzing, program analysis, exploit development, or research across complex software and infrastructure.
• Experience designing evaluations, benchmarks, or realistic environments for security tools and agents.
• A track record of original work through vulnerability disclosures, published research, open-source security tools, or independent projects.