Member of Technical Staff - Security Research

Runlayer; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Benefits offered by this job

Competitive salary

Equity

Paid time off

Professional development

Top-tier equipment

Health benefits

Customer exposure

Job summary

Runlayer in San Francisco seeks a Security Researcher to uncover vulnerabilities across MCP servers, AI coding agents, skills and plugins, including OAuth flows. You will disclose responsibly and publish findings that become protections for customers.

You will publish technical write-ups, develop open-source tools, and lead ecosystem-wide studies. Your work informs detections, scanner rules, and risk ratings for MCP security, while collaborating with marketing and developer-relations teams.

Qualifications

• 5+ years in offensive security research or red teaming.
• Public record: CVEs, advisories, talks or published tools/writeups.
• Depth in agent-native attacks (prompt injection, OAuth, supply-chain attacks).
• Proficient in Python, TypeScript or Go for harnesses, fuzzers and scanners.
• Clear writing for engineers and security leaders; strong disclosure judgment.

Responsibilities

• Find and exploit vulnerabilities in MCP servers, AI coding agents, and plugin ecosystems.
• Coordinate end-to-end disclosure: CVEs, advisories, embargoes, publication.
• Publish technical write-ups, open-source tools, and conference talks.
• Run ecosystem-wide studies across MCP servers using our catalog and scanning pipeline.
• Turn findings into product: detections, scanner rules and public risk ratings.
• Brief customers, prospects and press with our teams.
• Contribute to MCP specification security work and industry frameworks.

Skills

Offensive security

CVEs/Advisories

Agent-native attacks

Python/TypeScript/Go

Technical writing

Disclosure governance

AI-native security

Tools

Fuzzers

Scanners

Harnesses

Job description

About Runlayer

AI is transforming how every company operates, but most enterprises are stuck. They want to move fast with AI Agents, tools, and workflows, but they can't do it safely. We're fixing that.

Our team built AI Actions for OpenAI, shipped Zapier Agents to millions of users, and launched the first remote MCP server with Anthropic. We helped establish the protocol, and now we're building the platform enterprises need to actually put AI to work.

Runlayer is one platform for MCPs, Skills, and Agents: purpose-built security, fine-grained governance, and complete observability so organizations can go all-in on AI across the entire company without the risk. We just raised a $30M Series A led by Felicis, with participation from Khosla Ventures, bringing our total raised to $42M. Already trusted by Gusto, Instacart, Opendoor, dbt Labs, and Decagon.

About The Role

As our first Security Researcher, you'll find the vulnerabilities that define AI agent security and publish the research the industry reads. You'll hunt across MCP servers, AI coding agents, skills and plugins, and the OAuth flows that connect them. You'll disclose responsibly, and every finding becomes a protection our customers run.

Why You'll Thrive Here

• Impact: Your findings shape how enterprises, vendors and standards bodies think about agent security, and they ship as protections for our customers
• Excellence: Work with the team that helped establish MCP and a group of senior engineers from top security backgrounds
• Ownership: Own the research agenda end to end, from the first bug to disclosure, publication and the stage

What You'll Do

• Find and exploit vulnerabilities in MCP servers and clients, AI coding agents, agent frameworks, skills and plugin marketplaces, and the OAuth flows between them
• Run coordinated disclosure end to end: vendor contact, CVEs and advisories, embargoes and publication
• Publish research people quote: technical write-ups, open-source tools and conference talks
• Run ecosystem-scale studies across thousands of MCP servers using our catalog and scanning pipeline
• Turn findings into product: detections, scanner rules and public risk ratings for MCP servers
• Brief customers, prospects and press with our marketing and developer relations teams
• Bring what you find into MCP specification security work and industry frameworks

What We’re Looking For

• 5+ years in offensive security research, vulnerability research or red teaming
• A public record: CVEs or advisories, conference talks, or published tools and write-ups
• Depth in agent-native attacks: indirect prompt injection through tool output, tool poisoning, cross-server shadowing, confused deputies through OAuth, supply-chain attacks on skills and plugins
• Builder, not just breaker: you write Python, TypeScript or Go for harnesses, fuzzers and scanners.
• Clear writing for engineers and security leaders alike
• Sound disclosure judgment, including with vendors who push back
• AI-native: you use AI agents every day, as tools and as targets

Bonus Qualifications

• Published research on LLM, agent or MCP security
• Time on a security vendor's research team or at an offensive security consultancy
• Talks at Black Hat, DEF CON, RSA or similar
• Relationships with vendor security response teams and security press
• Open-source security tools with real users

What We Offer

We provide a competitive package designed to attract and retain top talent who can work effectively with enterprise customers.

• Competitive salary and equity — compensation that reflects your expertise and customer-facing responsibilities.
• Paid time off — paid vacation, paid sick leave, and paid parental leave.
• Professional development — budget for conferences, courses, and certifications in AI, enterprise software, and customer success.
• Top-tier equipment — your choice of laptop and accessories to create your ideal work environment.
• Health benefits — comprehensive health, dental, and vision coverage.
• Customer interaction opportunities — work directly with innovative companies and see the immediate impact of your work.