Principal AI Vulnerability Researcher

Company Overview

AIBound is pioneering AI security by providing the first comprehensive control plane designed for secure AI integration. The platform facilitates discovery, testing, and protection of every AI model, agent, and identity to identify risks in AI systems before they cause harm, enabling enterprises to innovate securely and rapidly. Founded by a former CISO of Palo Alto Networks and Workday, AIBound boasts a team of cybersecurity experts and is supported by leading investors, making it a key player at the convergence of AI innovation and cybersecurity.

Role Summary

As coding agents gain full privileges on endpoints—capable of accessing secrets, executing shell commands, modifying production branches, and using unreviewed third-party tools—AIBound’s Harness serves as the governing control plane. The Harness centrally manages policies and enforces them at endpoints via allow, ask, or deny mechanisms.

We seek a Principal AI Vulnerability Researcher to proactively identify and exploit vulnerabilities in this layer before malicious actors do, transforming these discoveries into defensive solutions. This is a foundational research role with hands-on responsibilities initially, reporting directly to the CEO, and offering the chance to expand and lead a research team as the function grows.

Key Responsibilities

• Conduct adversarial attacks on the Harness enforcement path, including policy delivery, tampering, endpoint policy files, decision hooks, and the gaps between policy rules and command executions.
• Identify and develop novel exploits targeting coding agents and the AI supply chain, such as prompt injection, manipulation of tool-call processes, context poisoning, and compromising MCP servers, skills, and sub-agents.
• Convert research findings into deployed protections via deterministic rules, decision hooks, and LLM-assisted classifiers, validated against labeled benchmark datasets prior to endpoint deployment.
• Keep abreast of the latest threats by translating research into product roadmaps, publishing technical write-ups, delivering talks, and contributing CVEs to establish AIBound’s authoritative research presence.
• Develop the research division by setting strategic agendas, recruiting, and mentoring team members as the function scales.

Candidate Qualifications

• Possess over six years of experience in offensive security, vulnerability research, exploit development, or detection engineering, including at least one year focused on AI, large language models, or agentic systems.
• Proven track record in original research evidenced by CVEs, advisories, conference presentations, bug bounty achievements, or creation of offensive tools.
• Deep understanding of macOS and Linux internals coupled with proficiency in shell command parsing and the ability to recognize command disguises and evasions.
• Advanced skills in Python and shell scripting with comfortable capability to write enforcement code, beyond just exploiting vulnerabilities.
• Sound working knowledge of agent system architectures, particularly in tool calling, MCP, skills, memory, and trust boundaries within these constructs.
• Strong commitment to responsible research practices, including sandbox usage and coordinated vulnerability disclosures.