Threat Research Engineer

An overview of this role

As an Threat Research Engineer at GitLab, you'll research emerging security threats and turn your findings into protection for GitLab customers. You'll investigate new attack surfaces, including AI coding agents, agent plugins, and package ecosystems, and build systems that help detect threats and improve vulnerability advisories. This is a full-stack role: you'll take an idea from the first research question to a working proof of concept, deploy it, and analyze the results yourself. Artificial intelligence (AI) is central to how we work. You'll use AI tools daily and build agentic workflows for research, analysis, and detection.

• Research new attack surfaces and threat vectors, including AI coding agents, agent plugins, and package ecosystems.
• Find and responsibly disclose vulnerabilities in open source projects and developer tools.
• Analyze software supply chain incidents and events.
• Build AI-powered workflows and agents that automate research, analysis, and detection.
• Build and maintain the cloud-based systems behind our detection and advisory work.
• Improve the quality and coverage of our vulnerability and malware advisories.
• Write up your findings as blog posts and present them at conferences.
• Hands-on security experience in research, malware analysis, penetration testing, or security tooling, with an understanding of package ecosystems and software supply chain attacks.
• Both a developer mindset and an adversarial mindset: you can build systems and think about how to break them.
• Demonstrated experience using AI in real work, such as building with large language model (LLM) application programming interfaces (APIs) or agent frameworks like LangChain, LangGraph, or agent software development kits (SDKs).
• Strong programming skills in Go or Python, with good software engineering practices, including testing, code review, continuous integration and continuous delivery (CI/CD), and writing maintainable code.
• Experience deploying and running production services on a public cloud.
• The ability to map attack vectors, explain trade-offs, and use data to move experiments forward, balancing quality, speed, and customer outcomes.
• Clear writing for engineers and a general audience, and keeping up with security and AI industry trends.
• Helpful, but not required: Google Cloud Platform (GCP); Common Vulnerabilities and Exposures (CVE) entries or published findings; research on AI agents, LLM applications, or Model Context Protocol (MCP) servers; blogs, talks, or open source security tools; Terraform or OpenTofu; databases or data pipelines at scale; reverse engineering.
• You don't need deep knowledge of compilers or static application security testing (SAST) engines. A high-level understanding is enough.

…