Senior Lead Specialist - Agentic AI (Cyber Defence & Resilience)

A hands-on Principal AI Engineer responsible for designing, building, and productionising secure agentic AI systems for cybersecurity use cases. The role will establish the agentic AI harness, control plane, model evaluation framework, system interfaces, memory architecture, guardrails, and observability standards, turning experimental AI agents and frontier models into safe, reusable, and measurable capabilities across cybersecurity functions.

AI Architecture & Infrastructure

• Design LLM agents with reasoning, planning, tool use, memory, retrieval, and task decomposition.
• Build the agentic AI harness, control plane, model routing, multi-agent coordination, and human approval workflows.
• Define autonomy levels, policy enforcement, approvals, and audit requirements.

System Integration & Tool Interface

• Integrate AI agents with enterprise cyber systems, including SIEM, SOAR, EDR, IAM, PAM, CMDB, ITSM, and vulnerability tools.
• Develop secure APIs, connectors, webhooks, queues, MCP interfaces, and service accounts.
• Implement scoped access, session controls, rate limits, and error handling.

Security & Access Control

• Establish clear permissions for agents to read, recommend, or execute actions.
• Implement identity controls, least privilege, just-in-time access, credential scoping, and secrets isolation.
• Secure prompts, tools, connectors, models, and other AI supply-chain components.

Knowledge & Data Engineering

• Build RAG, vector search, knowledge graphs, context stores, and case memory for cybersecurity workflows.
• Implement source-trust and evidence-provenance controls.
• Ensure outputs are grounded in reliable and traceable evidence.

Model Evaluation & Testing

• Evaluate frontier and open-source models for cybersecurity use cases.
• Develop evaluation frameworks, benchmarks, regression tests, adversarial tests, and simulations.
• Create secure cyber simulation and replay environments for agent testing.

AI Safety & Governance

• Protect agents against prompt injection, untrusted inputs, and unsafe actions.
• Define human decision rights, ownership, accountability, and approval requirements.
• Implement production monitoring, operational handover, LLMOps, and agent lifecycle management.

Deep AI/ML Expertise

• Strong hands-on experience building production-grade LLM, agentic AI, ML, automation, or platform systems.
• Deep knowledge of agent architecture, orchestration, tool calling, memory, RAG, model routing, and multi-agent workflows.
• Experience with frontier and/or open-source models, including model evaluation, benchmarking, and comparison.

Software Engineering

• Strong software engineering skills in Python, APIs, and backend services.
• Experience with cloud platforms, containers, CI/CD, authentication, logging, and observability.
• Ability to integrate AI systems with enterprise APIs, identity systems, and data platforms.

Production Systems

• Experience operating or supporting production systems, including monitoring, alerting, incident response, rollback, and release management.
• Understanding of AI production failure modes, including model drift, prompt regression, and failed tool calls.
• Experience with cost management, access controls, and post-incident reviews.

AI Safety & Risk Management

• Practical understanding of hallucination, prompt injection, insecure tool use, excessive agency, data leakage, memory poisoning, and adversarial manipulation.
• Experience designing human-in-the-loop controls for high-risk environments.
• Ability to support operational handover through clear runbooks, ownership, and support models.