Product Security Engineer

Traditional product security teams work one report at a time: a person triages a bug bounty submission, validates it, reproduces it, and hands it off for a fix. That doesn't scale past a certain volume, and Vercel is well past it. Adding more triagers doesn't close that gap. Building the systems that triage at that scale does.

This role is about building that system. Your core focus is tooling that triages and validates bug bounty and other externally reported security findings at scale, reasoning about validity, severity, and reproducibility the way a human triager would, but continuously and at volume. And we want to go beyond triage. The real leverage is in connecting a validated finding to its root cause and driving the fix, ideally with the remediation itself proposed or opened automatically for well-understood vulnerability classes.

More broadly, this is a mandate to rethink traditional security tooling for how Vercel actually operates: agent-scale testing and automation in place of processes built for a much smaller company. This role also has real scope to build tooling that gives our customers their own security testing capabilities for what they build on Vercel, not just harden Vercel's own surface.

Because of this, we're optimizing for someone who wants to build systems, not someone whose background is manual penetration testing. A software engineer with a strong desire to move into security, or a security engineer with a strong engineering background, is exactly who we're looking for.

If you're based within a pre-determined commuting distance of one of our offices (SF, NY, London, or Berlin), the role includes in-office anchor days on Monday, Tuesday, and Friday. If you're located beyond that distance, the role is fully remote. For location-specific details, please connect with our recruiting team.

• Build tooling to triage and validate bug bounty and external findings at scale: Design and operate the systems that take in externally reported vulnerabilities and automatically assess validity, severity, and reproducibility, at a volume no manual triage process could match.
• Push triage beyond pattern matching, into agentic analysis: Build and operate LLM/agent-based reasoning that can validate business logic, auth, and design-level findings, not just match against known signatures.
• Go from validated finding to root cause: Trace validated findings back to the underlying pattern or class, so the team fixes the reason it happened, not just the one report that came in.
• Build toward automated remediation, not just automated triage: Design systems that can propose, and increasingly open, the fix itself for well-understood vulnerability classes, with the right human review gates in place.
• Rethink traditional security tooling for scale: Question which parts of the traditional product security toolkit (manual threat modeling, ad hoc code review, point-in-time pentests) still make sense at Vercel's scale, and build the agent-driven tooling that replaces or augments them.
• Own and evolve the bug bounty program: Manage the researcher-facing side (scope, policy, engagement) as well as the internal tooling, so every report gets resolved and makes the automated triage smarter for the next one.
• Build toward customer-facing security testing capabilities: Extend the tooling and automation you build for Vercel's own products into a capability customers can use to test the security of what they build and deploy on the platform.