AI Security Architect

As the AI Security Architect, you will lead the security strategy, governance, risk management, and safe deployment of AI, LLM, ML, and agentic systems across platforms. You will define how AI systems are hardened against adversarial attacks, how AI infrastructure is secured end-to-end, and how AI governance frameworks are operationalised across engineering and product teams. Strong cybersecurity engineering experience is required; this role sits at the intersection of AI security and traditional cybersecurity, and the right candidate must bring depth in both. This role is part of a 0-to-1 vertical build. The person joining should be comfortable with ambiguity, high ownership, rapid iteration, hands-on execution, team building, and periods of high-intensity work during the early setup phase.
Responsibilities:
Define and own the enterprise AI security strategy covering LLMs, ML models, agentic systems, and AI infrastructure.
Conduct AI threat modelling: adversarial attacks, model inversion, data poisoning, prompt injection, model extraction.
Secure AI training and inference infrastructure: model serving hardening, API security, pipeline integrity.
Lead AI red-team initiatives: systematic adversarial testing of LLM and agentic systems in production.
Build AI governance frameworks: model risk management, provenance controls, audit trails, and approval workflows.
Secure agentic systems and orchestration layers: tool-use security, agent permission models, sandboxing strategies.
Define model integrity and provenance controls: signing, versioning, reproducibility, and tamper detection.
Conduct AI security assessments and vulnerability reviews across all production AI systems before and after deployment.
Establish secure AI deployment practices ensuring platform security across cloud infrastructure, containers, and API layers.
Implement privacy-preserving architectures (data masking, synthetic data, differential privacy) to safeguard corporate PII from leaking into LLM training sets or context windows.
Embed automated security compliance scanners, vulnerability checks, and input/output guardrails directly into the MLOps CI/CD pipelines.
Requirements:
Strong experience securing AI, ML, and LLM-based applications in production environments.
Deep understanding of AI threat modelling, prompt injection, jailbreaks, data poisoning, and adversarial attacks.
Strong knowledge of OWASP LLM Top 10 and AI security best practices.
Experience conducting AI security assessments, red teaming, and vulnerability reviews.
Understanding of model governance and secure AI deployment.
Strong hands-on cybersecurity experience across cloud, application, and platform security, including penetration testing, vulnerability management, SIEM/SOC experience, or security architecture at the infrastructure layer.
AI Ecosystem Knowledge is a must.
Conceptual clarity on transformer architectures, fine-tuning pipelines (SFT/RLHF), and vector search mechanics.

Experience
5-9 yrs