Senior AI Security Consultant

Government Technology Agency; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

Government Technology Agency seeks a Senior AI Security Consultant to architect secure AI workflows within a government data hub. You will ensure auditable, non-repudiable AI actions across multi-agency data sources while maintaining public trust.

You will collaborate in a Discovery Squad with AI engineers to balance rapid prototyping and security rigor, focusing on policy-as-code, immutable logging, and privacy-preserving techniques.

Qualifications

• 6–10 years of cybersecurity experience with 2–3 years in AI/ML or securing LLM apps.
• Proven track record in cloud infrastructure security: serverless, API gateways, cloud databases.

Responsibilities

• Security Architecture & Threat Modeling for autonomous AI systems.
• Design workflows with human-in-the-loop approvals for high-stakes decisions.
• Build auditable chains of custody for AI decisions and data access.
• Ensure reproducibility and data governance across agency data sources.
• Conduct proactive security testing and apply advanced privacy measures.

Skills

Adversarial mindset

Risk communication

Collaboration in high-velocity env

Tools

ABAC (Policy-as-Code)

Immutable logging

Job description

As the Senior AI Security Consultant for the BusinessSG initiative, you will be the primary security architect guiding our transition to aproactive AI-driven platform. You will be embedded within a 10-man Discovery Squad, working alongside AI engineers and product leads to ensure that autonomous AI workflows—operating via messaging interfaces and a unified data hub—are secure, transparent, and resilient.

Your mission is to ensure every AI action is auditable and non-repudiable, protecting multi-agency data sources while maintaining the highest standards of public trust.

Key Responsibilities

1. Security Architecture & Threat Modeling
   • System Design: Lead the design of security guardrails for AI agents to prevent them from taking unauthorized legal or financial actions.
   • Threat Modeling: Map out potential risks for autonomous AI, such as users trying to "hijack" the agent's intent or tricking it into accessing data it shouldn't.
   • Human-in-the-Loop: Design workflows that ensure high-stakes AI decisions (like grant recommendations) cannot proceed without explicit, authenticated human approval.
2. Reliable Audit & Explainability
   • Audit Trails: Build a "Chain of Custody" for AI decisions by logging what the agent "thought," what specific data it retrieved, and which model version was used.
   • Reproducibility: Ensure that AI actions are documented well enough to be reconstructed and explained to auditors or stakeholders.
   • Data Governance: Ensure the AI only accesses integrated agency data on a strict "need-to-know" basis.
3. Testing & Advanced Privacy (Preferred)
   • Security Testing: Conduct proactive testing to find vulnerabilities in how the AI processes instructions or uses external tools.
   • Advanced Privacy: Implement specialized privacy technologies to protect sensitive business information within our unified data hub.

Required Competencies & Experience

Experience Profile

• 6–10 years of experience in Cybersecurity, with at least 2–3 years focused on AI/ML systems or securing LLM-based applications.
• Proven track record in Cloud Infrastructure Security, specifically securing serverless compute, API gateways, and cloud-native databases.

Core Technical Foundations

• Agentic System Design: Experience architecting or securing systems where AI moves beyond "chat" to performing actions through APIs.
• LLM Security Foundations: Strong understanding of current LLM/ASI security risks, including prompt injection and data privacy in RAG (Retrieval-Augmented Generation) setups.
• Regulatory Knowledge: In-depth knowledge of Singapore’s public sector security standards, including:
  • IM8.
  • Model AI Governance Framework.

Strategic & Soft Skills

• Discovery Squad Collaboration: Ability to thrive in a high-velocity "sandbox" environment where you must balance rapid prototyping with security rigor.
• Adversarial Mindset: Ability to anticipate how users might try to "game" the system to gain unfair advantages.
• Risk Communication: Translating complex technical AI risks into clear, actionable advice for senior leadership.

Preferred Qualifications

• Experience with Policy-as-Code or logic-based authorization systems (e.g., ABAC).
• Knowledge in Digital Forensics or building immutable logging architectures.
• Prior experience in a Public Sector or "Smart Nation" project context.

Why this role?

This is a unique opportunity to join a specialized squad at the ground floor of agentic AI adoption in government. You will move from defending static applications to securing autonomous intelligence that directly impacts the Singapore business community.