OpenAI is hiring: Forward Deployed Engineer (FDE), Cyber (Tokyo)
OpenAI's Forward Deployed Engineering team is hiring a cyber security engineer in Tokyo to work with customers' security teams, putting AI agents to work on patching and attack detection. 5+ years of software engineering and hands-on AI agent work required.
The role
OpenAI's Forward Deployed Engineering (FDE) team builds with customers to put frontier AI into production. Its Cyber engineers work with the security teams of organisations that run critical systems, for example patching hundreds of thousands of repositories or sifting petabytes of logs for signs of malicious agents. With the customer's engineers you design the system, wire AI agents into security workflows and decide which actions need a human sign-off, then build evaluations (does a patch stop a known exploit without breaking the app?) and take it to production. Where today's models fall short, you take the gap back to OpenAI's Research and Product teams and help close it.
What they ask for
- 5+ years as a software engineer, shipping production software
- A background keeping production systems safe: security engineering, infrastructure, SRE, DevOps or product engineering
- Has built and evaluated AI agents for software engineering work; knows CI/CD well
- Can find their way in an unfamiliar codebase and land maintainable changes
- Explains technical trade-offs clearly to engineers and security leaders
Nice to have
- Application security (CodeQL or Semgrep rules, fuzzing) or tools that turn security findings into tested patches
- Detection engineering with Sigma, YARA, Splunk or KQL; cloud and identity security
- Malware or exploit research; testing AI agents against prompt injection and tool misuse
Good to know
- Location: Tokyo, Japan; hybrid, three days a week in the office, up to 50% travel. OpenAI offers relocation help.
- Posted on 20 November 2026, according to OpenAI's job data.
- OpenAI says it mainly wants strong engineering generalists; deep security experience is a plus.
How the interview usually goes
For customer-facing engineering roles like FDE, interviews usually lean towards practical coding, such as working in an existing codebase or writing a small tool, more than pure algorithms, plus a talk about how you work with customers. Because this is an AI job, expect extra questions on LLM and machine-learning basics, how agents use tools and how prompt injection attacks them, recent models or research, evaluating an agent, and designing a production AI system.
