See how candidates think and approach the work this role demands, before the phone screen. We'll build a video challenge from this posting, and you can edit or share it before it goes live.
An AI search company is moving into cybersecurity by selling agents that perform offensive and defensive security work for customers, signaling a bet that autonomous agents can replace parts of the security workflow rather than just assist analysts.
ABOUT THE ROLE
Perplexity is building cybersecurity products where agents do the security work, not just assist with it. Our security team already runs agents on our own environment every day, across finding vulnerabilities, testing our defenses, securing our cloud, and responding to threats. Now we're turning that work into products for customers, and we expect those products to span application security, offensive testing, cloud security, detection and response, and areas we haven't picked yet.
We're looking for a hands-on Software Engineer to build these products end to end, from the security logic and agent workflows to the integrations and product surfaces customers use. You'll ship to design partners and enterprise customers quickly, and you'll work directly with the security engineers who do this work internally, so what we build is grounded in real codebases, real environments, and real attackers.
RESPONSIBILITIES
Build security products from zero to one across domains such as application security, penetration testing and offensive security, cloud security, and detection and response.
Build agents that do real security work: review code and find vulnerabilities, test applications and infrastructure like an attacker, assess cloud configurations, and investigate threats, then validate findings and recommend or apply fixes.
Build the systems underneath: multi-tenant services, data pipelines, and secure execution environments that run against customer code, cloud accounts, and telemetry.
Build integrations with the tools customers already use: source control and CI/CD, cloud providers, identity providers, security tools, ticketing, and chat.
Ship the product surfaces customers use: APIs, dashboards, onboarding, and workflows that fit how security and engineering teams actually work.
Hold the product to the same standard we hold ourselves: strong tenant isolation, least privilege, safe handling of customer secrets and data, and tight controls on any agent that acts in a customer environment.
Define how we measure quality (finding accuracy, false positive rate, exploitability, time to fix, cost) and use it to drive improvements.
Work directly with design partners and customers, and help decide which security problems we should build products for next.
QUALIFICATIONS
4+ years of experience as a software engineer building production systems, with meaningful time spent on security products, security data platforms, or security engineering with heavy software development.
Proficiency in at least one major programming language (such as Python, Go, TypeScript, or Rust) and experience owning production services end to end.
Hands-on depth in at least one security domain, such as application security, penetration testing or offensive security, cloud security, or detection and response, and curiosity to work across the others.
Experience integrating with security- or developer-relevant systems such as source control, CI/CD, cloud provider APIs, identity providers, scanners, or SIEMs.
Familiarity with cloud infrastructure (AWS preferred) and building secure, multi-tenant SaaS.
Comfortable working directly with customers and shipping pragmatic improvements quickly in an ambiguous, early-stage product area.
Bonus: experience building LLM-powered agents, evaluation frameworks, or tool-using automation for security or other high-stakes domains.
Bonus: experience at a security product company (AppSec, pentesting, cloud security, detection and response) or on an internal security platform at scale.
perplexity:b23d83e3-add8-48ea-b16e-65a956d1192cSee how we measure for definitions, or our corrections log for known issues. Found something wrong? Flag a correction.
perplexityThis posting hasn't been probed by our closure verifier yet. Stream C runs on a rolling schedule against postings approaching the close-decision threshold.