AI Security Researcher

Cogent Security • ,

11d

About The Position

Cogent is hiring a Security Researcher to build the offensive side of its AI security research: the attacker that Cogent’s defensive agents must beat. The role sits in Research and owns the process of turning real red-team tradecraft into autonomous AI capabilities. The researcher will build and improve red-team agents, expand them toward full-chain operations across enterprise environments, develop and validate evasion capabilities, operate realistic cyber ranges, and create benchmarks to measure agent performance. The role also includes validating critical and high-severity attack paths in customer environments and representing Cogent in the offensive security community through research, talks, and open-source contributions.

Cogent is an Applied AI Lab developing AI agents for cybersecurity. Its research team combines frontier research with real-world execution and partners with Fortune 500 companies to secure complex production environments. Cogent is backed by Greylock and Bain Capital Ventures.

Requirements

• At least 6 years of experience in offensive security, red-team operations, or adversary emulation.
• Hands-on experience running full-chain operations end-to-end, from initial access through post-exploitation, across Windows, Active Directory, and on-premises environments.
• Depth in Command and Control frameworks, such as Mythic, and payload development.
• Real detection-evasion experience against EDR.
• Ability to write code, such as Python, to turn manual tradecraft into repeatable, automatable capabilities.
• Experience standing up and operating realistic test environments or cyber ranges with production defenses such as EDR, SIEM, and logging.

Nice To Haves

• Cloud attack experience involving identity and access management, misconfiguration, or CI/CD, or network attack experience.
• Experience building evaluations or benchmarks that measure security capability.
• Open-source tooling, conference talks, or published research in offensive security.
• Experience working with AI agents, LLM tooling, or agentic workflows.

Responsibilities

• Design and build red-team agents, including their agent harness, tools, Model Context Protocol (MCP) servers, and skills that encode operator workflows from scoping and reconnaissance through payload development, delivery, Command and Control, and post-exploitation.
• Review agent runs and address capability gaps.
• Extend red-team agents from application and code security into full attack chains, including initial access, delivery and evasion, Command and Control, privilege escalation, and lateral movement across Windows, Active Directory, and on-premises environments.
• Work with cloud security researchers to extend coverage into cloud identity and access management, misconfiguration, continuous integration and delivery, and network security.
• Build detection-evasion knowledge into agents and the test harness, and operate detonation and detection loops that validate capabilities against real Endpoint Detection and Response (EDR) systems.
• Design and operate realistic cyber ranges with defenses such as EDR, Security Information and Event Management (SIEM), and logging, initially emphasizing Windows and Active Directory and later extending to cloud and hybrid environments.
• Convert attack paths into evaluation tasks with verifiable success criteria and develop benchmarks, including feeding work into IntrusionBench, to measure agent improvement.
• Verify critical and high-severity attack paths in customer environments and feed findings back into agents and detections.
• Represent Cogent in the offensive security community through research, talks, and open-source contributions.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: No Education Listed