Senior AI Cybersecurity Engineer

About the role

You will build agentic offensive security tools: AI agents that find, validate and exploit vulnerabilities in applications, source code, cloud environments and infrastructure, the way a skilled red teamer would.
You are first and foremost a strong software engineer. You also know how attackers work, well enough to teach an agent to think like one. And you have built agentic applications before, so you know what it takes to make them reliable in production.
This is an applied AI role. You will build on top of large language models, not train them.

 

What you will do

• Design and build production agentic systems that carry out security testing autonomously, covering planning, tool use, memory and multi-agent orchestration.

• Turn attacker tradecraft into agent capability: translate attack techniques into tools, prompts, workflows and decision logic.

• Build the tooling agents rely on, such as exploit harnesses, sandboxed execution environments, browser and API automation, and code analysis.

• Build evaluation harnesses and benchmarks that measure how well agents find real vulnerabilities, how many false positives they raise, and what each run costs.

• Reproduce and triage what the agents report, then feed the lessons back into the system.

• Build in the controls that keep agents safe to run: target scoping, guardrails, human approval points and audit logging.

• Ship and operate the software end to end: services, APIs, user interfaces, CI/CD, tests and observability.

• Use AI coding assistants and agents in your own daily work, and raise the team's practice in doing so.

• Lead technical design, review code and mentor other engineers.

• Keep current with new attack techniques and with developments in LLMs and agent tooling.

 

What we are looking for

Software engineering

• At least 5 years of professional software engineering experience, building and operating production systems.

• Strong in Python and TypeScript/JavaScript.

• Solid grounding in backend services, APIs, data stores, containers and CI/CD, on at least one major cloud platform.

• Good engineering habits: automated testing, code review, clear system design and documentation.

Agentic AI

• Hands-on experience building and shipping at least one agentic or LLM-powered application.

• Practical command of the building blocks: tool calling, context management, multi-step orchestration, structured outputs and agent protocols such as MCP.

• Experience evaluating LLM systems, including writing evals, iterating on prompts and managing cost and latency.

• Awareness of how these systems fail, including hallucination, non-determinism and prompt injection, and how to design around it.

• Routine use of AI tools in your own engineering workflow.

Offensive security

• Working knowledge of attack techniques across all four domains below, with hands-on experience in at least two.

• Web and API: OWASP Top 10, authentication and authorisation flaws, injection, SSRF and business logic abuse.

• Whitebox testing: reading source code to find vulnerabilities, trace them to an entry point and prove they are exploitable.

• Cloud: IAM misconfiguration, privilege escalation and service abuse on AWS, Azure or GCP.

• Infrastructure and red team operations: network attacks, Active Directory, lateral movement, command and control, and the MITRE ATT&CK framework.

• Ability to write proof-of-concept exploits and explain findings clearly to engineers.

 

Good to have

• Time spent on a penetration testing or red team.

• Offensive security certifications such as OSCP or OSWE.

• A track record in CTFs, bug bounty programmes or published CVEs.

• Contributions to open-source security or agent tooling.

• Experience securing AI systems themselves, such as agent sandboxing and prompt injection defence.

• Front-end experience with a modern framework such as React

Singaporeans only.